A Systematic Mapping Study on Automatic Control Systems of Multi-Port dc/dc Power Converters

Abstract

In the ongoing transition to renewable energy sources, power converters have become indispensable. Their prevalence is increasing, enabling efficient energy conversion, enhancing reliability and stability, and optimizing power extraction from renewable sources. Multi-port dc/dc power converters are widely used because they offer advantages in managing multiple sources and loads. However, designing an automatic control system for these converters presents a challenge due to their complexity. Many configurations for multi-port dc/dc power converters have been proposed, featuring diverse combinations of controllers, modulation techniques, and topologies tailored to specific applications. The body of knowledge on these configurations has grown. Yet, papers have been published according to the authors’ areas of specialization, thus generating a scattered and unorganized body of knowledge and making it difficult to discern research trends and open challenges. Previous studies have attempted to organize knowledge about these configurations, but they have not established a systematic mapping process that follows a rigorous and objective methodology. This paper conducts a systematic mapping study on Automatic Control Systems of multi-port dc/dc power converters. Our study analyzed 122 papers from the 777 papers found around the topic to find and organize the body of knowledge on topology, controller, efficiency, number of elements, modulation technique, and practical applications. This systematic mapping provides a foundational framework for researchers, aiming to inspire further exploration and the development of innovative controller systems in multi-port dc/dc power converters. We found the application of machine learning techniques in dc/dc power converters constitutes an open challenge in these devices.

1. Introduction

Renewable energy sources emerge as viable solutions to manage energy deficits; however, they are not continuous or predictable for constant use. The combination of multiple energy sources contributes to ensuring a steady electrical supply. Various renewable sources, including solar, wind, biomass, and tidal energy, are being studied continuously. It requires the incorporation of at least one power converter per energy source. Additionally, power converters play a critical role in Distributed Generation, which involves small-scale electricity sources connected directly to distribution networks or user-side meters, offering improved power quality and reliability compared with traditional solutions. As noted in [1], Distributed Generation provides higher power quality or reliability than conventional solutions. Moreover, in transmission systems, the integration of HVDC (high voltage direct current) grids, as discussed by Agbemuko et al. [2], reduces costs and improves overall efficiency.

Modern dc/dc power converters are engineered to provide efficient energy conversion while ensuring stability, safety, robustness, and well-regulated dc voltage. The term “dc/dc” refers to power converters that manage direct current at both the input and the output ports. Traditionally, conversion is achieved in a single stage, where the source and load are directly connected through a power converter. Hence, a multi-sourced complex system requires various synchronized power converters. Each power converter fulfills a specific function, such as collecting energy, managing energy storage, compensating imbalances between generation and consumption, delivering energy to users, or performing voltage conversion, among other tasks. Multi-port power converters are specifically designed for these multi-sourced systems, integrating multiple conversions into a single stage. These converters are popular because of their benefits: high power density, reduced power losses, fewer filters, smaller size, improved efficiency, and lower costs compared with traditional converters. Multi-port power converters find applications in electric/hybrid vehicles, telecommunications, satellites, Uninterruptible Power Supply (UPS), and more.

New proposals for multi-port converters continuously emerge, each offering improvements by introducing variations in topology, control systems, or modulation techniques tailored for specific applications, as documented in [3,4,5]. The reported configurations could potentially be valuable in other applications, simplifying the development process. In addition, an organized database would facilitate the creation of new configurations and help in selecting the best features from each proposal. Despite extensive research published on this topic, there is a lack of a comprehensive summary that encompasses all relevant characteristics, topologies, control systems, applications, element quantities, and achieved efficiencies. Furthermore, there is no documented basis to identify gaps in the existing research on power converter structures.

A systematic mapping study provides a broad overview of existing evidence in a research domain, especially when the topic is wide or evidence is scarce. Unlike systematic reviews, it maps studies at a high level to identify evidence clusters and gaps, guiding future research and systematic reviews [6]. This paper presents a systematic mapping study of automatic controllers tested and reported on multi-port dc/dc power converters. The study aims to explore recent advances in multi-port dc/dc power converters from both topology and control perspectives, providing a valuable reference for future converter design and applications. The study considers various factors, including controller algorithms, modulation techniques, topologies, efficiencies, applications, and the number of elements. The goal is to identify current limitations and outline future research directions in this domain to structure a research agenda. Data was collected from the Scopus database, covering the period from 2005 to 2023, following the procedure proposed by Petersen et al. in [7].

The following sections are structured as follows: Section 2, Background, includes a conceptual framework for understanding the classification scheme used. Section 3 discusses the contributions reported in related work. Section 4 details the methodology of the systematic mapping study, based on general guidelines for conducting systematic mapping study (SMS) studies in software engineering proposed by Petersen et al. [7]. Section 5 presents the findings of the mapping study. Section 6 discusses the main discoveries. It presents the research agenda, including identified research topics and recommendations for future research directions. Section 7 examines potential threats to the validity of the study. Finally, Section 8 presents the conclusions of the paper.

2. Background

This section presents a conceptual framework to understand and support the classification scheme used. First, we described power converters and their types, specifically focusing on dc/dc multi-port power converters.

The function of power converters is to transform electrical energy from one form (voltage, current, or frequency) to another. This ensures compatibility between the source and load, facilitating the transfer of electrical energy between different components. Depending on the form of input and output waveforms, power converters are classified as dc-to-dc (dc/dc) converters, inverters (dc-to-ac), rectifiers (ac-to-dc), and cyclo converters (ac-to-ac). The energy can be harnessed from a variety of sources, including solar panels, fuel cells, regenerative braking systems, and mechanical vibrations. They are used in a diverse range of electronic systems, including communication relay stations, traffic signals, sensor networks, hybrid vehicles, laptops, and handheld devices.

Multi-port converters represent a significant advancement in the field of power conversion. By integrating multiple conversion functions within a single device, they enable the flow of power between ports. This capability is of critical importance for systems that require connection to multiple sources, including but not limited to multiple wind turbines, photovoltaic systems, modular structures, and hybrid electric vehicles. As discussed by Sato et al. [8], power converters can be complex and expensive as they comprise multiple power sources. Multi-port converters address the complexities and cost considerations associated with managing multiple power sources, with a particular emphasis on efficient power management and grid integration, as highlighted by Jianwu Zeng et al. [9].

In converter design, the objective is to achieve the maximum possible efficiency, with researchers aiming for levels approaching 100%. Kumar et al. [10] highlight that the utilization of multi-port converters with a reduced number of devices results in enhanced efficiency in comparison to those with a greater number of devices. The efficiency of a converter is primarily determined by its topology, which refers to the arrangement of electrical components such as switches, diodes, capacitors, and inductors. The selection of an appropriate topology is of critical importance and is dependent upon the specific requirements of the intended application and the desired level of efficiency. The most common dc/dc converter topologies include buck, boost, buck-boost, Cuk, Sepik, bridge, forward, and flyback configurations [11]. These are classified into isolated and non-isolated types based on electrical connectivity between input and output, as outlined in reference [12].

Power Semiconductor switches, such as metal-oxide semiconductor field-effect transistors (MOSFETs) and insulated-gate bipolar transistors (IGBTs), are integral to the construction of converters. Their conduction states are managed effectively through the use of modulation techniques. Modulation techniques, including Pulse Width Modulation (PWM), Frequency Modulation (FM), Phase Shift Modulation (PSM), Amplitude Modulation (AM), and Space Vector Modulation (SVM), as described by Almutairi [13], ensure efficient converter operation by regulating output parameters while minimizing harmonic distortions.

3. Related Work

Although there have been previous works reporting about multi-port dc/dc power converters, either they do not cover all the characteristics we have analyzed or do not established a relationship between them. Also, the majority were not developed systematically, which does not help their replicatio;, the number of papers in this work is 123. However, they are complementary to ours and are useful for consolidating the growing knowledge base of multi-port dc/dc power converters. In the following paragraphs, we analyze these previous works and compare our approaches and results with theirs.

In the last twelve years, an increasing number of configurations for dc/dc power converters have been proposed. Gevorkov et al. [14] published a review and analysis of the most significant aspects of multi-port converters, considering various characteristics, topologies, benefits, drawbacks, and application areas. Still, it does not provide a cross-reference between the control system and the converter architecture. Rehman et al. [15] present trends in developing multi-input and multi-output dc/dc converters and provide a general comparison of multi-input non-isolated dc/dc converter topologies, highlighting their limitations. However, they do not consider other characteristics such as modulation techniques, applications, or the number of elements.

Hossain et al. [16] conducted a comprehensive review of power dc/dc converters, focusing on topologies in various applications and medium-voltage dc power systems. The study also provides an overview of modulation techniques and the state-of-the-art control strategies of well-established converters. However, this study has some limitations: it is not related to multi-port power converters, it does not consider the number of elements in a power converter or its efficiency, and it primarily focuses on the characteristics of the converter without analyzing recent work in the field.

Similarly, Kolli et al. [17] give a comprehensive overview of the power converter interfaces potentially favorable for the automotive, railways aircraft, and small stationary domains. This review article highlights the importance of choosing the right power converter architecture and the related technology. Furthermore, the work presented in [18] reviews the state-of-the-art research works on non-isolated dc/dc buck, boost, buck-boost, Cúk, and Sepic converters and their characteristics to find a solution best suiting an application with Maximum Power Point Tracking in photovoltaic applications. Both are focused on a limited application and do not consider additional characteristics in this work.

Additionally, a notable work that, although it does not encompass all aspects considered in this study, provides practical conclusions about specific types of power converters. Forouzesh et al. [19] comprehensively review and classify various step-up dc/dc converters based on their characteristics and voltage-boosting techniques. They discuss the advantages and disadvantages of these techniques and the associated converters in detail. Finally, the paper presents and summarizes broad applications of dc–dc converters, including a comparative study of different voltage-boosting techniques.

Dhananjaya et al. [20] provide an overview based on the literature on MISO, SIMO, and MIMO converters, detailing different schematic configurations, switching pulse signals, gain, and the number of elements in various converters. The paper also discusses challenges and future research directions. In contrast, the present work records all modulation techniques and controllers reported in the literature, along with a detailed characterization of the converters, to help identify areas for future investigation.

Babaei et al. [21] present various types of multi-port dc/dc converters, discussing operating modes, advantages, disadvantages, schematic circuits, the number of inductors, capacitors, switches, and output voltage equations for each configuration. In contrast, the present work focuses on the automatic control of dc/dc multi-port power converters, reporting on the number of devices, topology, efficiency, and applications.

Finally, Hao Zhang et al. [22] propose a systematic review and analysis of the most significant aspects of multi-port converters. It includes classifications based on various characteristics, topologies, benefits, drawbacks, and areas of application. Differences with the present study include a different classification scheme is used, focusing mainly on topology, and control systems are not considered in the classification scheme, though it mentions three types: Proportional–Integral controller (PI), Proportional–Integral Derivative controller (PID), and Model Predictive Voltage Controller (MPVC). Additionally, the present work reports general applications without specifying the types of converters.

4. Methodology

This study presents a Systematic Mapping Study (SMS), a methodical approach aimed at providing a comprehensive overview of automatic control systems for multi-port power dc/dc converters. The process involves three distinct stages: planning, conducting, and reporting.

In the planning phase, the research objectives and research questions (RQs) were defined, and Selection Literature Mapping was planned. This planning encloses the paper selection strategy, inclusion and exclusion criteria, the classification scheme, the extraction process, and the identification of visualization tools.

The execution stage involves executing the SMS protocol established during the planning phase. This phase began with the selection of scientific databases and the extraction of paper data. Then, the paper selection was developed through an iterative process based on inclusion/exclusion criteria and validated through peer review. The selected papers are then organized using the classification scheme developed in the planning phase. This systematic categorization facilitates both single- and cross-dimensional analyses, aiding in the identification of gaps and trends.

Finally, in the reporting stage, the findings of the study were analyzed using the data gathered during the conducting stage. This comprehensive analysis provides a general vision of automatic control systems for power dc/dc converters, as was planned through the systematic mapping study.

4.1. Scope of Study

This study focuses on a characteristic of a multi-port dc/dc power converter. Figure 1 illustrates the scope of the study, which encompasses the intersection of two domains: control schemes and multi-port dc/dc power converters.

The domain of control schemes involves automatic control systems, which consist of two defined parts: Modulation Techniques, which concern signal control applied to power semiconductors for their conducting or non-conducting modes, and the Control System, which relates to control algorithms employed in closed-loop configurations for multi-port dc/dc power converters.

The domain of multi-port dc/dc power converters allows for power converters capable of handling multiple energy sources, with each port being either unidirectional or bidirectional. Diverse configuration topologies of these converters are registered for discussion in the study.

The objectives of this study are summarized as follows:

• To summarize the state-of-the-art automatic control systems applied to multi-port dc/dc power converters;
• To provide a suitable base to outline gaps and trends in the research field and, subsequently, lead the way for potential research to contribute to filling them in.

A Systematic Mapping Study promotes knowledge about the state of the art of automatic control systems applied to multi-port dc/dc power converters; it is a basis for identifying gaps, trends, and possible research directions in this area. To achieve objectives following research questions (RQs) are formulated:

RQ1: What are the automatic control systems applied to multi-port dc/dc power converters?

Control strategies play a crucial role in maximizing the efficiency of multi-port converters. This is because control techniques can enhance the overall operations of these converter topologies [14]. Additionally, the controller, modulation technique, and simplicity of design are other significant factors in operating a converter at its peak efficiency [23]. The control strategy provides signals to control power switches based on the input signal obtained from sensors, enabling control over the output signal. It is responsible for various characteristics and performance aspects of the power converter. Among the directly affected variables are reliability, robustness, response time, quality, harmonic content, power output, voltage level, and current level, among others. The control system effectively manages the primary actions of the power converter.

RQ2: What are the different topologies used in a multi-port dc/dc power converter for reported automatic control systems?

Non-isolated topologies transfer power without magnetic isolation. They do not use a transformer and lack the advantages of galvanic isolation, such as a high step-up voltage gain ratio; instead, they have simpler configurations and do not suffer from disadvantages of galvanic isolation, such as magnetic interference or high weight. These characteristics make them suitable when size and weight are important concerns in particular applications. The voltage gain of isolated topology is generally higher than its non-isolated counterpart. However, the transformer design procedure and alleviating the leakage inductance effect are important factors in these converters [14].

RQ3: How efficient are multi-port dc/dc power converters with different control systems?

The efficiency of the power converter facilitates the evaluation of the effectiveness in achieving the maximum amount. It indicates the extent of energy loss during the conversion process. Elevated heating of the power switches leads to reduced efficiency. This information aids in identifying optimal solutions within the field and provides a comprehensive overview of software and hardware options for dc/dc power converters.

RQ4: What are the reported applications of multi-port dc/dc power converters?

A wide range of voltage conversion ratios is frequently required across numerous applications, including renewable energy systems like batteries and photovoltaic modules. In scenarios where galvanic isolation is not necessary, non-isolated dc/dc converters can be employed to accomplish voltage step-up or step-down, resulting in a reduction in size, weight, and volume. This reduction is attributed to increased efficiency due to the absence of a high-frequency transformer [24]. Applications demand specific characteristics for dc/dc power converters; hence, it is necessary to distinguish each area and prioritize to reach high efficiency, ensuring stability, safety, robustness, and well-regulated dc voltage.

RQ5: How many elements (passive and semiconductors) does the converter have?

Developing a prototype with fewer elements is cheaper during the implementation stage. Fewer components typically lead to lower manufacturing and maintenance expenses. Additionally, devices generate energy losses, which decrease the efficiency of the power converter. Furthermore, controlling more semiconductors necessitates more complex algorithms because these devices must work synchronously. Simplifying design can often lead to increased reliability and easier troubleshooting, enhancing overall system robustness. Finally, a simplified power converter design can promote advancements in controllers and the topology of power converter technology.

4.2. Paper Selection Strategy

In this study, a database-based strategy is employed, using the Scopus database to find high-quality refereed research literature, comprising both journal and conference papers. Scopus indexes high-quality peer-reviewed papers from major digital libraries pertinent to the field of power electronics, such as IEEE Xplore, Springer Link, Science Direct, and ACM. Moreover, this database has been referenced in other pertinent papers that have conducted related systematic studies.

The investigation employed the Systematic Mapping Study as its base. We constructed a search string that incorporates vocabularies and taxonomies relevant to the research domain. This search string was formulated by combining research domains, control schemes, and multi-port dc/dc power converters. To ensure the retrieval of as many papers as possible, the search encompassed top-level terms like “control of multi-port dc/dc power converters,” along with synonyms of controllers derived from taxonomies and vocabularies such as the 2017 IEEE Thesaurus Version 1.0.

Table 1. Search string.

((‘control W/2 system’ OR ‘control’)

AND (‘multi-port’ OR multiport OR mpt)

AND (‘power AND converter’ OR ‘dc W/1 dc’))

displays the final search string alongside its associated domains and terms. This search string underwent an iterative process, wherein the outcomes of each iteration were validated using a test set of 10 pertinent papers supplied by an experienced researcher in power electronics. As a result of the search in the Scopus database, 777 papers were identified and progressed to the next stage, which involved the inclusion and exclusion procedure.

4.3. Inclusion and Exclusion Procedure

The inclusion and exclusion procedure is depicted in Figure 2. Although an automated procedure on Scopus is not available, the use of other software, such as Cadima, facilitated the detection and filtering of duplicate referenced papers.

The manual processing involved screening based on titles, abstracts, and full texts to include reported primary paper contributions within the domain of interest. This study utilized the CADIMA tool for paper selection, illustrated in Figure 3. Therefore, the criteria for including papers are outlined as follows:

• Papers focused on multi-port dc/dc power converters.
• Papers detailing control systems for the converters.
• Papers constituting primary contributions.

Cadima identified 24 duplicated papers, which were afterward removed from the database. To ensure inter-coder reliability, each stage underwent a pilot phase conducted by the researcher and an experienced professor until a kappa coefficient exceeding 0.80 was achieved. In the screening stage, each paper could be labeled as either Included, Excluded, or Unclear.

For each paper, the main screening consisted of three stages: screening based on title, abstract, and full text. Before each stage, a pilot phase was conducted where the screener and tutor worked with different sets of papers provided by Cadima automatically until a kappa coefficient greater than 0.80 was attained.

During the title-based screening, out of 753 papers, a pool of 437 candidate papers was labeled as either included or unclear and progressed to the next stage. Similarly, in the abstract-based screening, 381 candidate papers were labeled as included or unclear and advanced to the final stage.

In the full-text screening, screeners conducted a comprehensive examination of the title, abstract, introduction, and conclusions. If insufficient information was available, the screeners checked the section where the controller design was explained. At this stage, 381 papers were reviewed by one screener. Papers that were not clearly defined were labeled as unclear. Finally, both screeners discussed and determined the status of each paper. As a result of this stage, 122 papers were included, and 259 were excluded.

Figure 3. Manual criteria for inclusion.

Figure 3. Manual criteria for inclusion.

4.4. Classification Scheme

The classification scheme presented is based on available theory and scientific works about multi-port dc/dc power converters. As depicted in Figure 4, this scheme outlines five key research questions addressed in this study. The classification aims to delineate the primary characteristics related to the design, implementation, efficiency, and application of multi-port dc/dc power converters to facilitate replication and performance enhancement.

Control strategies enclose both the control system and modulation technique. The control system commands the behavior of the power converter. There are two types of control systems commonly used: open-loop and closed-loop systems, Multi-port dc/dc power converters do not employ an open-loop control system; therefore, this paper does not address them.

A closed-loop control system regulates the output signals from the power converter (typically current, voltage, or power) based on reference values and measured signals, usually the output signal itself. It furnishes a control signal that governs the operation of power switches through a modulation signal. Control theory presents various control systems that have been applied in power converters, including classical controllers, optimal control, robust control, sliding mode control (SMC), predictive control, nonlinear control systems, and control systems with artificial intelligence (AI).

Classical controllers aim to achieve stability and desired performance based on a known mathematical system model, utilizing traditional control methods like P, PI, and PID controllers. Optimal control optimizes a performance index over time, looking for the best control signals to achieve specific objectives while requiring the system model. Robust control ensures the operation of the system in the presence of uncertain parameters or disturbances despite modeling errors and uncertainties in the system. SMC maintains system stability by forcing the state of the system to slide along a predefined surface in the state space. Predictive control uses a dynamic system model to predict future behavior and optimize control actions for desired performance. Nonlinear control systems address modeling errors and unmodeled dynamics, achieving high-accuracy state estimation and effective control, often employing model-based filtering approaches and input–output feedback linearization (IOFL) techniques.

Finally, Control Systems with artificial intelligence (AI) incorporate techniques such as neural networks, fuzzy logic, reinforcement learning, and genetic algorithms to learn from data, adapt to changing conditions, and optimize control actions. It is important to note that control systems can also be categorized according to whether they depend on a mathematical model or not.

In the context of multi-port dc/dc power converters, while there is no specific classification available, Sinha et al. (2018) [25] provide a classification of modulation techniques for grid-connected multilevel inverters (GCMLIs), which can also be applicable to our purposes. Their classification broadly categorizes modulation techniques into high-switching frequency, fundamental switching frequency, and variable switching frequency.

High-frequency modulation techniques offer efficient power transfer, reduced components size and weight, enhanced transient response, and minimized electromagnetic interference. Moreover, they enable higher power densities and contribute to system stability. Two primary methods in this technique are Space Vector Pulse Width Modulation (SVPWM) and Sinusoidal Pulse Width Modulation (SPWM). SVPWM involves comparing a complex voltage vector control signal with a reference vector to obtain the output voltage. SPWM works by comparing a sinusoidal reference signal with a carrier waveform, typically a high-frequency triangular or sawtooth waveform. SPWM modulation can be classified as carrier-based or reference-based. Carrier-based SPWM can further be categorized as single-carrier (generic PWM) and multi-carrier, where the latter is classified based on the number of carriers, including Phase Shift PWM (PS-PWM) and left-shifted PWM (LS-PWM). In PS-PWM, the PWM signals are generated with varying phase shifts relative to each other, reducing harmonic distortions. However, LS-PWM adjusts the level or magnitude of PWM signals to ensure effective driving of power devices, minimize switching losses, and improve efficiency. LS-PWM can be further classified into Phase Disposition PWM (PDPWM), Phase Opposition Disposition PWM (PODPWM), Alternative Phase Opposition Disposition PWM (APODPWM), and Multi-Carrier Sinusoidal PWM with variable frequency.

Fundamental switching frequency PWM refers to a modulation technique in which the switching frequency of the PWM signal is kept low, typically within the range of the fundamental frequency of the output waveform, such as the mains frequency (50 or 60 Hz) for grid inverters. Operating at low switching frequencies can reduce switching losses and electromagnetic interference (EMI), as fewer switching events occur per unit of time compared with higher switching frequencies. Several low-frequency modulation techniques are presented, including the selective harmonic elimination (SHE) technique, Nearest Vector Control (NVC), Nearest Level Control (NLC), also known as space vector control (SVC), and Nearest Level Modulation (NLM).

Variable switching frequency PWM dynamically adjusts the switching frequency according to certain operating conditions or requirements, helping to improve overall system efficiency, reduce losses, and enhance transient response. In addition, other general modulation strategies, such as hybrid modulation strategy and hysteresis PWM, offer combinations that enhance performance and extreme robustness, greater stability, and faster dynamic response compared with other PWM techniques.

The main goal of the multi-port converter is to integrate multiple power input nodes into a single device while allowing power transfer between each node. Various classifications based on topology have been proposed by different authors. We categorize dc/dc power converters into isolated and non-isolated multi-port configurations, primarily based on their isolation characteristics. The distinguishing criterion is whether the converter offers electrical isolation between the input and output stages.

Multi-port dc/dc power converters are gaining popularity due to their versatility. Although the proposed classification schemes consider a limited number of applications, it is essential to note that dc/dc power converters are applicable across all areas where power conversion is required. Efficiency serves as the primary quantitative metric to determine which converter maximizes energy utilization in a given application. Various configurations have been tested, and our emphasis is on measuring efficiency and understanding the components used in the construction of each converter. The large number of switches, diodes, and inductors employed in the converter increases the size, weight, and cost of the multi-port dc/dc power converter; it is more representative of a large number of inputs.

Finally, a large number of switches, diodes, and inductors increases the size, weight, and costs of the multi-port dc/dc power converter. This effect is especially critical when there are many ports (inputs/outputs).

5. Results

A total of 753 related papers were identified and collected from Scopus. Of these, 437 articles were selected for title-based screening, and 381 articles underwent abstract-based screening. Finally, 122 articles were selected through a comprehensive screening process, and all are listed in Appendix A. Figure 5 shows the number of articles related to the research area published since 2005, highlighting a significant growth rate starting in 2014 and peaking in 2019. This trend may be related to the increasing effects of global warming observed over the years.

Before the codification process, the data were tabulated and analyzed to address the research questions, as detailed below:

5.1. RQ1: What Are the Automatic Control Systems Applied to Multi-Port dc/dc Power Converters?

The paper compiles the automatic control systems reported for multi-port dc/dc power converters. Inspired by the classification presented in [26], this study systematically groups controllers into categories, as shown in

Table 2. Types of controllers identified in multi-port dc/dc power converters.

Category	Controller	N	ID
Classical	PI [55], PID [6], Voltage control [2], current control [1], P [1].	65	[ID343, ID286, ID519, ID64, ID201, ID236, ID299, ID334, ID395, ID436, ID535, ID570, ID312, ID488, ID503, ID592, ID129, ID135, ID144, ID213, ID419, ID427, ID478, ID32, ID250, ID260, ID371, ID420, ID448, ID140, ID505, ID745, ID197, ID628, ID719, ID179, ID252, ID281, ID439, ID491, ID637, ID641, ID172, ID13, ID27, ID151, ID253, ID340, ID363, ID373, ID387, ID392, ID594, ID692, ID179], [ID514, ID160, ID739, ID337, ID605, ID109], [ID20, ID418], [ID424], [ID370].
Modern	Fuzzy [4], LQR [1], ADPM [1], Multi-objective [1], Multivariable [1], Time-multiplexed [1], PPP [1].	10	[ID627, ID265, ID88, ID391], [ID377], [ID633], [ID356], [ID180], [ID467], [ID2].
Hybrid	PI+SMC [1], predict-reset PI [1], PI+HESS [1], P&O MPPT [5], PS-PWM [1].	9	[ID521], [ID626], [ID310], [ID96, ID98, ID103, ID666, ID703], [ID518].
Predictive	MPC [1], NLMPC [1], PEAA [1].	3	[ID93], [ID614], [ID16]
Intelligent	PI+ANN	1	[ID330]
PV Algorithm	MPPT-based control schemes	18	ID251, ID431, ID738, ID372, ID741, ID421, ID722, ID124, ID589, ID678, ID441, ID765, ID565, ID4, ID723, ID766, ID682, ID711
Open loop		5	ID530, ID694, ID759, ID688, ID545
Not reported		11	ID452, ID184, ID254, ID564, ID365, ID274, ID522, ID298, ID506, ID207, ID263

. Control systems are categorized by complexity, with sixty-five papers classified as classical, ten as modern, nine as hybrid, four as predictive or intelligent, and eighteen related to algorithm-based PV approaches, ensuring comprehensive coverage. Furthermore, 16 papers do not explicitly report the type of controller and, therefore, have not been considered in this classification.

Classical controllers are the most frequently reported in the literature, accounting for approximately 65% of the studies. Specifically, the Proportional–Integral (PI) controller is the most commonly applied control strategy in multi-port dc/dc power converters between 2010 and 2022, being used in around 52% of the cases. PI controllers are typically chosen due to their simplicity, reliability, and ease of implementation in digital control systems, especially when testing new topologies in closed-loop configurations [27]. They offer effective steady-state error elimination [28], low computational cost [29], high reliability, and robustness while reducing system order and aligning well with the ON–OFF behavior of power electronic switches [30]. However, it should be noted that a reported limitation is their reduced efficiency in high-voltage, large-scale photovoltaic (PV) applications [31].

Proportional–Integral–Derivative (PID) control is another commonly reported strategy in multi-port dc/dc power converters, accounting for approximately 6% of the reported cases between 2010 and 2022. Unlike the PI controller, the PID controller includes a derivative component that anticipates future error trends by evaluating the rate of change. This results in a damping effect that improves the transient response of the system and reduces overshoot. Consequently, PID controllers are often regarded as effective in balancing speed, stability, and accuracy in control applications. However, they are less frequently implemented than PI controllers, primarily due to the increased complexity and computational resources required to tune all three parameters, which can outweigh their benefits in many practical scenarios. In the reported studies, PI controllers have consistently demonstrated satisfactory performance with simpler implementation requirements. However, both PI and PID controllers require appropriate tuning methodologies and typically depend on an accurate mathematical model of the system. Their performance under transient conditions may be limited, particularly in highly dynamic or nonlinear environments. To overcome these limitations, hybrid approaches that integrate classical control with modern techniques, such as predictive, advanced, and intelligent controllers, have been proposed and investigated in the literature. In addition, voltage and current controllers (reported in approximately 3% of studies from 2019 to 2022) are commonly used due to their simplicity and intuitive design. These controllers offer good transient response to load variations and input voltage fluctuations, making them suitable for dynamic operating conditions. They operate by implementing closed-loop regulation of a reference voltage or current, which modulates the duty cycle to control the switching of power devices. Some reported implementations, such as ID20, incorporate dual-loop control schemes that combine voltage and current feedback to improve overall system performance.

Modern controllers account for approximately 9% of the reported studies between 2015 and 2022. In ID377, the authors recommend the use of the Linear Quadratic Regulator (LQR) method as a more effective approach to determine the gain matrix. LQR is a model-based control strategy designed to minimize a cost function that reflects the performance of the system over time. The multi-objective control algorithm described in ID356 is developed to manage battery operation, direct power flow, select the mode of operation, and determine the appropriate duty cycle in a non-isolated high-gain triple-port dc/dc buck–boost converter. The mode selection and corresponding switching signals are determined on the basis of current photovoltaic (PV) power, the state of charge (SOC) of the battery or its preset current limit, and the load demand. A simple voltage control scheme combined with a Maximum Power Point Tracking (MPPT) technique is used to maintain the output voltage. Additionally, the Partial Power Processing (PPP) scheme proposed in ID2 is based on predictive control. It reduces DC-link current, volume, and power losses by ensuring that most of the energy transfer between low-voltage ports is handled by only a portion of the current in the solid-state transformer, thus improving efficiency.

Hybrid control systems are also reported in the literature, accounting for approximately 8% of the studies published between 2012 and 2022. These systems combine classical control strategies with advanced or intelligent techniques to enhance performance under complex operating conditions. Among the hybrid approaches identified are those PI controllers with Artificial Neural Networks (ANNs), classified under intelligent control. Additionally, Sliding Mode Control (SMC), a robust technique capable of driving the system state to a desired condition and maintaining it despite model uncertainties or external disturbances, is commonly used in nonlinear environments. Another notable approach is consensus-based control, a decentralized strategy applied in multi-agent systems, where multiple units (e.g., sensors or subsystems) cooperate to achieve a unified control objective. These techniques are particularly suited for managing nonlinearities and estimating parameter variations in dynamic systems. However, the implementation of such hybrid controllers often demands significant computational resources and advanced real-time processing capabilities.

The Model Predictive Control accounts for approximately 3% of the reported studies between 2015 and 2022. This category includes the nonlinear Model Predictive Controller (NMPC) and the Predictive Energy Amendment Algorithm (PEAA). In ID93, the general formulation for NMPC is presented, along with the C/GMRES algorithm, which enables efficient real-time implementation. The PEAA is designed to optimize energy flow within a system, improving efficiency, reliability, and sustainability. This is achieved by forecasting the availability of energy from multiple sources and estimating future demand based on historical consumption patterns, real-time data, and predictive modeling. However, these controllers are heavily relying on accurate mathematical models, which can limit the robustness of the system in the presence of modeling errors or unforeseen disturbances.

Photovoltaic (PV) systems involve several interconnected processes, including energy harvesting, storage, transmission, and conversion. Numerous studies address control strategies targeting each of these subsystems. However, this review focuses specifically on the primary control strategy highlighted in each publication. While MPPT (Maximum Power Point Tracking) algorithms are not traditional feedback controllers like PI or PID, they are frequently responsible for generating reference values for inner-loop controllers. Moreover, in many PV-related studies, the MPPT algorithm is the only control mechanism reported. For this reason, MPPT is included in our classification, as it plays a central role in the overall control strategy of multi-port DC-DC power converters, particularly in renewable energy applications. The Maximum Power Point Tracking (MPPT) algorithm—reported in approximately 17% of the reviewed studies between 2006 and 2022—is widely used in power converters to maximize solar energy extraction and improve overall system efficiency. MPPT algorithms adjust the duty cycle of the converter to match the source impedance with the load impedance, thereby achieving maximum power output. Some of the most commonly implemented MPPT techniques include Perturb and Observe (P&O), Incremental Conductance (INC), Fractional Short Circuit Current, Fractional Open Circuit Voltage, Artificial Neural Networks (ANNs), and fuzzy logic. Kumar et al. [32] present a comparative analysis of these techniques in tabular format, enabling evaluation of their performance and assisting in the selection of the most suitable method for specific applications. Additional comprehensive reviews of MPPT techniques can be found in [33,34].

The modulation strategies identified in the literature were ranked according to their frequency of use, as summarized in

Table 3. Types of modulation techniques identified in multi-port dc/dc power converters.

Category	Techniques	N°	ID
PWM	PWM	60	ID592, ID286, ID334, ID312, ID431, ID518, ID213, ID377, ID738, ID514, ID160, ID250, ID370, ID371, ID372, ID420, ID739, ID741, ID759, ID96, ID140, ID337, ID421, ID614, ID722, ID745, ID124, ID197, ID356, ID627, ID688, ID98, ID179, ID252, ID265, ID281, ID439, ID441, ID467, ID506, ID666, ID703, ID765, ID109, ID263, ID545, ID565, ID694, ID2, ID16, ID20, ID27, ID151, ID310, ID363, ID392, ID594, ID692, ID723, ID766.
PWM Variants	PWM+VSFPWM, PWM + PPAS, PWM + PFM, PWM+PSM, PSC-PWM, Optimized PWM, special PWM switching, Saturation of the i-th duty cycle	21	[ID207], [ID103], [ID448], [ID605], [ID343, ID519, ID299, ID535, ID570, ID503, ID522, ID589, ID682, ID711, ID637, ID13, ID340, ID424], [ID452], [ID641], [ID628].
Phase-shift/Specialized Modulation	PSC, PPAS, QPSM, EPS, Q2L	17	[ID184, ID93, ID236, ID251, ID436, ID530, ID488, ID135, ID365, ID419, ID478, ID260, ID678], [ID719], [ID395], [ID254], [ID373].
Frequency-based/Switching	PFM, VF-PWM, SWM, SPSC	5	[ID491, ID180], [ID298], [ID418], [ID129].
Space Vector Modulation (SVM)	SVPWM, 3D - SVM	2	[ID88], [ID177].
Other/Unreported	Pulse Generator, Not Reported	17	[ID201], [ID64, ID564, ID144, ID427, ID32, ID274, ID505, ID633, ID330, ID521, ID626, ID172, ID4, ID253, ID387, ID391].

. Based on this analysis, modulation techniques were grouped into categories.

Pulse-width modulation (PWM) accounted for approximately 49% of total reports between 2006 and 2023, making it the most widely adopted technique. PWM is favored for its simplicity, ease of implementation, and compatibility with digital control platforms. However, it suffers from drawbacks such as attenuation of the fundamental frequency amplitude and increased total harmonic distortion (THD). Although increasing the switching frequency can mitigate THD, it also increases switching losses, imposes additional stress on the devices, and introduces high-frequency harmonics [35]. Several studies highlight adaptations of PWM to address these issues. For example, ID207 and ID298 implement variable-frequency PWM to enable boundary current mode operation, optimizing inductor sizing and improving efficiency. ID641 proposes a specialized PWM scheme for dual-input operations, while ID452 presents an optimized PWM method designed to minimize conduction losses through fundamental frequency analysis of current and power signals.

The PWM variants represent 17% of the studies (2012–2022). Notably, Phase-Shift Carrier PWM (PSC-PWM) is widely reported because of its ability to merge the benefits of both PWM and PSC techniques. ID260 presents a modified phase-shift modulation scheme for forward power flow control in full-bridge converters. ID254 describes three modulation strategies: a PSC-PWM variant for Modular Multilevel Converters (MMCs), a quasi-square wave modulation method, and the Extended Phase Shift (EPS) control scheme, which adapts to varying system conditions. Similarly, ID719 introduces a pulse width modulation with phase angle shift (PWM-PPAS) strategy, allowing independent regulation of the duty cycle and phase angle for decoupled port voltage control. ID103 further develops this idea, proposing a soft-switching control technique that enhances the decoupling and device-sharing capability across i/o ports.

Article ID719 discusses a control scheme referred to as pulse width modulation plus phase angle shift (PPAS), where the switch duty cycle and the phase angle of interleaved converters are used as two control freedoms to achieve decoupled voltage regulation within a certain operating range among different ports. In ID103, a control scheme is introduced to achieve soft switching, decoupled control, and improved device-sharing capability across inputs. This scheme is referred to as the PWM phase angle shift (PWM-PPAS) control scheme, enabling zero voltage switching for switches.

Phase-shift and specialized modulation techniques account for approximately 14% of the reviewed literature (2012–2022). PSC stands out for its advantages in reducing electromagnetic interference, improving efficiency, and lowering switching losses and THD. In ID395, Quintuple Phase Shift Modulation (QPSM) is proposed to improve soft-switching and reduce circulating current under dc-link mismatches. The Quasi Two-Level (Q2L) technique, featured in ID373, maintains a near-two-level waveform with reduced switching losses and EMI, supporting modular topologies and simultaneous multi-port power flow—key in scalable systems such as hybrid storage and renewable energy.

Frequency-based switching techniques account for 4% of the literature (2018–2022). These include pulse frequency modulation (PFM), which, as reported in ID491, improves efficiency under light loads. Variable frequency PWM (VF-PWM), employed in ID298, adapts the switching frequency to ensure zero voltage switching (ZVS). ID418 utilizes Single-Wavelength Modulation (SWM) to enhance harmonic performance and simplify digital control. ID129 introduces Soft Pulse-Shifted Control (SPSC), which optimizes switching transitions in modular converters and ensures voltage balancing across submodules, demonstrating the effectiveness of hybrid strategies for high-density EV charging applications.

Space Vector Modulation techniques represent 2% of the reviewed works (2021–2022). Among these, the 3D Space Vector Modulation (3D-SVM) method, aligned with industrial standards, provides enhanced flexibility in current control. Space Vector PWM (SVPWM) identifies and sequences the nearest three voltage vectors to construct the reference vector $V_{ref}$, optimizing the switching process.

Lastly, 14% of the studies did not explicitly report on the modulation technique used. This lack of information highlights a recurring document gap and suggests the need for improved reporting standards in future publications.

Figure 6 presents various combinations of control strategies published in the literature on multi-port dc/dc power converters. A significant majority of these works implement Pulse Width Modulation (PWM) in conjunction with classical control strategies, particularly the Proportional–Integral (PI) controller, which remains the most frequently applied due to its simplicity, ease of implementation in digital control platforms, and satisfactory performance in steady-state conditions. However, PI controllers struggle to track non-DC signals without steady-state error and often require careful tuning to maintain performance under varying operating conditions. To address these limitations, several studies adopt enhanced PWM techniques such as PSC-PWM, PFM, and VF-PWM, which enable soft switching, reduce electromagnetic interference, and improve converter efficiency under dynamic loads (e.g., ID641, ID298, ID491).

For improved dynamic performance, Pole Placement Control has been introduced to set desired poles based on state-space equations. Moreover, Linear Quadratic Regulator (LQR) control (e.g., ID377) offers a robust solution for achieving finite-time steady-state behavior and minimizing energy consumption during transient states. Model Predictive Control (MPC) and its nonlinear variant (NMPC), as in ID93 and ID614, are gaining attention for their ability to optimize cost functions under multivariable constraints; however, their reliance on system modeling and variable switching frequencies introduces complexity in filter design.

In addition, intelligent and hybrid controllers are emerging as promising alternatives. For example, combinations such as PI + ANN (ID330), PI + SMC (ID521), and predict-reset PI (ID626) aim to improve adaptability and robustness in highly nonlinear environments. These are often accompanied by advanced modulation techniques like SPSC and adaptive phase-shift strategies (ID129, ID103), which contribute to improved power sharing and voltage balancing in modular architectures.

Modern nonlinear control techniques also appear in the literature. Hysteresis control, while offering simplicity and robustness, is hindered by variable switching frequencies. Sliding Mode Control (SMC) and feedback linearization (FBL) provide systematic frameworks for improving system robustness and handling uncertainties. Adaptive control schemes dynamically adjust controller parameters in response to system variations, offering stability and performance across a wide operating range. Furthermore, the integration of artificial intelligence (AI) techniques is becoming more prominent, leveraging learning algorithms to improve control precision and system autonomy in complex energy applications [36].

Finally, the literature reflects a growing tendency to combine traditional control approaches with modulation strategies tailored to application-specific demands, whether focused on switching loss reduction, dynamic load tracking, or control robustness. This convergence of modulation and control design represents a shift toward more adaptive and intelligent power conversion architectures, especially in applications such as renewable energy integration, nanogrids, and EV fast charging systems.

5.2. RQ2: What Are the Different Topologies Used in a Multi-Port dc/dc Power Converter for REPORTED Automatic Control Systems?

Figure 7 illustrates the relationship between control strategies (horizontal axis) and modulation techniques (vertical axis) in reported multi-port dc/dc converter systems, highlighting the prevailing configurations and integration trends in automatic control design. Among the 122 papers reviewed, approximately 57% implement isolated topologies (reported between 2006 and 2022), while 43% employ non-isolated designs (reported between 2010 and 2023).

A total of 69 studies employ isolated converter architectures, which are often selected for their galvanic isolation, modular design, and suitability for renewable energy integration applications. The most prevalent modulation technique in these systems is conventional PWM, used in approximately 35% of isolated implementations (e.g., ID213, ID250, ID420). Isolated systems also adopt a diverse range of specialized modulation strategies: 16% of isolated converters report the use of PSC-PWM (e.g., ID711, ID682, ID589), which enables soft switching and reduced EMI. Other notable techniques include saturation of the i-th duty cycle (ID628), a fine-tuned method for precise power sharing; QPSM (ID395); SPSC (ID129); PFM (ID491); and Q2L (ID373), reflecting a growing interest in efficiency-enhancing soft-switching strategies. Additional strategies such as EPS (ID254), PWM-PPAS (ID103), and PWM+PSM (ID605) enable decoupled control and Zero-Voltage Switching (ZVS) across multiple power paths. Some studies (e.g., ID452, ID207) explore optimized or variable-frequency PWM to support boundary conduction modes and harmonic mitigation.

PI controllers dominate isolated converter control strategies, cited in more than 49% of cases (e.g., ID140, ID363, ID420), and are typically paired with PWM due to their simplicity and ease of digital implementation. Other classical controllers include PID (ID605, ID739), voltage control (ID418), and current control (ID424). For photovoltaic applications, MPPT algorithms are frequently integrated, particularly in renewable energy contexts (e.g., ID421, ID722, ID765), often in conjunction with PI controllers (e.g., ID711, ID682). Fewer studies explore modern or intelligent control approaches in isolated systems, with a few exceptions such as NMPC (ID93), PEAA (ID16), and fuzzy logic (ID627, ID88).

Isolated converters appear to facilitate more advanced modulation strategies, likely due to their modular and port-separated architecture. However, the control side remains heavily biased toward classical PI/PID strategies, suggesting a potential gap in the adoption of modern control techniques in isolated systems.

Fifty-three studies implement non-isolated converters, which are generally favored in cost-sensitive or space-constrained applications, such as consumer electronics, microgrids, or battery balancing systems. PWM also dominates the modulation landscape in these systems and is reported in more than 67% of non-isolated implementations (e.g., ID286, ID334, ID688). Although PSC-PWM appears in several cases (e.g., ID13, ID340, ID522), advanced modulation strategies are less prevalent than in isolated designs. Some notable exceptions include VF-PWM (ID298), special PWM switching (ID641), PFM (ID491), PPAS (ID719), and 3D-SVM (ID177), which suggest minor experimentation with hybrid switching methods.

Although PI controllers are still widely used, appearing in 37% of non-isolated designs (e.g., ID27, ID197, ID594), the diversity of control strategies in these systems is greater than in isolated configurations. PI control variants have been introduced to improve response time, robustness, and adaptability. Examples include PI+HESS (ID310), predict-reset PI (ID626), and PI+SMC (ID521), along with AI-based approaches (ID330). To manage nonlinearities, some systems employ MPC (ID614). Other reported methods, though less common, include fuzzy logic (ID265, ID391), multivariable control (ID180), multi-objective optimization (ID356), PPP (ID2), and time-multiplexed strategies (ID467). Conventional controllers are still reported as well, including P (ID370), PID (ID109, ID160, ID337, ID514), voltage control (ID20), and current control (ID177).

Non-isolated topologies offer a more fertile ground for control innovation, even though modulation techniques remain relatively conventional. The broader range of control algorithms in these systems reflects their suitability for data-driven, AI-enhanced, and predictive approaches, given their simpler physical constraints. In contrast, isolated topologies, with physical decoupling through transformers, tend to favor innovation in modulation techniques. However, a lack of systematic integration between control and modulation strategies is evident in both topologies. Many designs continue to default to PI-PWM combinations, even in scenarios where more advanced alternatives may yield improved performance. The underutilization of hybrid approaches in isolated systems, despite their modularity and application in renewable energy contexts, suggests an opportunity for further research to integrate intelligent and predictive control methods in such platforms.

5.3. RQ3: How Efficient Are Multi-Port dc/dc Power Converters with Different Automatic Control Systems?

An analysis was performed to evaluate the relationship between efficiency and output power in multi-port dc/dc power converters, considering the modulation technique, controller type, and converter topology.

In

Table 4. Efficiency and power analysis by modulation technique and topology.

Modulation Technique	Topology	Studies (N)	Efficiency Range	Power Range	Data Completeness
PWM	Isolated	24	80–98%	10 W–3000 W	46% Efficiency, 29% Power
PWM	Non-Isolated	36	68–99.5%	48 W–3200 W	50% Efficiency, 38% Power
PWM Variants	Isolated	20	89–99%	100 W–1500 W	35% Efficiency, 35% Power
PWM Variants	Non-Isolated	3	93.20%	100 W–2000 W	33% Efficiency, 66% Power
Others *	Isolated	19	84–96%	400 W–3240 W	42% Efficiency, 42% Power
Others **	Non-Isolated	5	86–98.5%	500 W–3900 W	40% Efficiency, 60% Power

* Includes: SPSC, PSC, Pulse Generator, EPS, Q2L, QPSM, PFM, and saturation of the i-th duty cycle. ** Includes: 3D-SVM, PFM, and PPAS.

, an analysis of the modulation techniques and converter topologies is presented based on the reported efficiency and output power. The data completeness metric indicates the percentage of studies that provide explicit information about both efficiency and power. In isolated topologies, the most frequently used modulation technique is PWM, often combined with control strategies to enhance performance. The highest efficiency is observed in entry ID421, where basic PWM modulation achieves 98% efficiency at 100 W, likely due to an optimized converter design and minimal control overhead. Other notable combinations include PWM with fuzzy logic (ID627), reporting 90% at 90 W, and PWM with PEAA, achieving 96.2% efficiency, highlighting the advantages of adaptive predictive control. Furthermore, PWM paired with PI and PID controllers yields peak efficiencies of 96.77% and 94%, respectively, which can be attributed to the improved steady-state accuracy and dynamic response of closed-loop controllers.

In non-isolated topologies, combinations such as Fuzzy + PWM (ID265) achieve 94.3% at 125 W, while MPPT + PWM (ID441) in PV applications reaches 95% at 120 W, benefiting from real-time adaptation to environmental conditions. Furthermore, PWM combined with PI control (ID252) reaches a peak of 98.2% at 100 W, which emphasizes the precision of the controller in output regulation. Notable high-power applications include PWM with voltage mode control (ID20), reporting 97.7% at 1600 W, which is likely due to the stability and fast transient response of this control strategy. Other significant results include PWM + P&O MPPT (ID98), which achieves 97.48%, and multi-objective optimization + PWM (ID356), resulting in 93.6%, both of which illustrate the benefits of integrating intelligent and adaptive control methods.

Regarding PWM variants, PSC-PWM combined with PI control (ID299) achieves 96% at 100 W, and PSC-PWM with MPPT (ID589) reaches 97%, confirming that modulation schemes with reduced switching losses and adaptive control provide efficiency gains. QPSM + PI (ID395) also shows a strong efficiency performance of 96%, while the use of VSFPWM achieves an outstanding performance of 99%, underscoring the effectiveness of adaptive frequency control in minimizing switching and conduction losses. In the non-isolated topology, PSC-PWM (ID522) reports a respectable 93.2% efficiency at 100 W, although generally lower than its isolated counterparts, due to higher parasitic losses.

For other modulation techniques, open-loop PSC in isolated configurations shows the lowest efficiency (84%), indicating the importance of feedback for performance optimization. In contrast, PSC with PI (ID260) reaches 96% at 3kW, illustrating the value of closed-loop control, particularly at high power levels. SPSC with PI reports the efficiency at 97%, although without specifying the output power, making direct comparison difficult. In non-isolated systems, PPAS with PI (ID719) reaches a maximum of 90%, while 3D-SVM with current loop control achieves 96% at 200 W, reflecting the benefit of advanced Space Vector Modulation in combination with current-mode control for precise and efficient operation.

In general, combinations that integrate PWM or its variants with closed-loop controllers, such as PI, MPPT, or intelligent control algorithms, consistently demonstrate higher efficiency. Non-isolated converters often achieve higher efficiencies when paired with advanced modulation techniques, largely because of the absence of galvanic isolation losses and their typically superior electromagnetic interference (EMI) characteristics under optimized control.

Table 5. Efficiency and power analysis by controller and topology.

Controller	Topology	Studies (N)	Efficiency Range	Power Range	Data Completeness
Classical	Isolated	38	86–98.5%	100 W–3240 W	34% Efficiency, 36% Power
Classical	Non-Isolated	27	68–99.5%	2.25 W–2000 W	48% Efficiency, 48% Power
Hybrid	Isolated	7	86–93%	10 W–500 W	33% Efficiency, 33% Power
Hybrid	Non-Isolated	3	79.3–97.48%	-	25% Efficiency, 0% Power
Modern	Isolated	3	90%	90 W	33% Efficiency, 33% Power
Modern	Non-Isolated	7	82.7–94.3%	125 W–250 W	28% Efficiency, 28% Power
Predictive	Isolated	1	-	1000 W	0% Efficiency, 100% Power
Predictive	Non-Isolated	3	97–99.5%	3300 W	33% Efficiency, 33% Power
PV applications	Isolated	12	85–97%	100 W–3900 W	66% Efficiency, 41% Power
PV applications	Non-Isolated	3	93–99.5%	150 W–32,000 W	100% Efficiency, 100% Power

presents the relationship between the controller type, the converter topology, and the reported efficiency. For isolated topologies using classical controllers, current control combined with PSC-PWM modulation is reported (ID424), delivering 100 W output; however, no efficiency data are provided. The PI controller appears in various combinations with different modulation techniques: PI + PSC-PWM achieves up to 96% efficiency at 105 W (ID299) and 90% efficiency at 500 W (ID637); PI + QPSM (ID395) reaches 96% efficiency at 100 W; PI + PSC (ID260) achieves 96% at 300 W and 90% at 3.24 kW (ID135); PI + PWM shows a peak efficiency of 96.77% at 100 W; PI + pulse generator reaches 90.5% at 1.2 kW; and PI + PWM-PFM achieves 95% at 150 W. In non-isolated topologies, the PI controller is the most frequently reported. PI + PWM yields efficiencies ranging from 68% at 200 W to 98.2% at 100 W (e.g., ID27, ID252, ID312). PI + PSC-PWM (ID13) supports up to 2 kW but does not report efficiency, while a similar case appears in ID641, where PI is combined with a special PWM switching method. PID + PWM (ID109) achieves 90–94% efficiency at 100 W, indicating that PI-based schemes generally provide better efficiency. The use of current loop control with 3D-SVM modulation shows 95–96% efficiency at 200 W, suggesting that the modulation technique enhances performance. PWM + voltage control (ID20) reports an efficiency of 97.7–97.9% at 1.6 kW.

For hybrid controllers, PI + MPPT (ID682) achieves 89–93% efficiency at 100 W to 500 W, while PO + MPPT (ID666) reports 86–92% at 10 W to 100 W.

Among modern controllers, only one isolated topology case (ID627) reports an efficiency of 90% at 90 W with a fuzzy + PWM configuration. In non-isolated topologies, Fuzzy + PWM (ID265) reaches 94.3% efficiency at 125 W, while multi-objective control schemes range between 82.7% and 93.6%.

Predictive controllers offer simple implementations with good dynamic performance. Although efficiency and power are not consistently reported, they improve the transient response. For example, MPS with PSC modulation supports 1 kW (ID93), and Predictive Reset PI achieves 97–99.5% efficiency with a load of 3.3 kW. A PI controller enhanced with an Artificial Neural Network (ANN) is also proposed, though neither efficiency nor output power if specified. However, ANN contributes to improved robustness.

In PV applications, MPPT combined with PWM controllers reaches between 85% at 1 kW (ID766) and 98% at 100 W (ID421). MPPT with PSC improves efficiency to 90–93% at 400 W, and ID4 reports 96% at 3.9 kW, although no modulation technique is specified. In non-isolated topologies, MPPT + PWM (ID565) achieves 98% efficiency at 32 kW. In general, non-isolated topologies generally achieve higher average efficiencies than isolated ones. PWM-based techniques, particularly PSC-PWM and optimized PWM, are dominant across a broad power range and deliver consistently high efficiencies. PSC-based modulation is widely used in isolated systems and maintains good efficiency, often exceeding 90%.

Figure 8 presents a comparison of the highest recorded efficiencies for different controller strategies implemented in multi-port dc/dc power converters. Both isolated and non-isolated topologies are included. The predict-reset PI controller with a non-isolated topology at 3.3 kW. It is notable that non-isolated topologies generally exhibit higher efficiencies compared with their isolated counterparts. This performance advantage can be attributed to the absence of high-frequency transformers in non-isolated designs, which reduces core and switching losses. Additionally, non-isolated converters often have simpler circuit structures and fewer power conversion stages, minimizing conduction losses and improving overall energy transfer efficiency. The superior performance of the predict-reset PI controller highlights the potential of enhancing traditional PI-based schemes. This controller is a modification of the conventional PI regulator, integrating predictive and reset actions that help reduce overshoot and steady-state error, thereby improving dynamic response and energy efficiency. While conventional controllers such as PI and PID remain widely used due to their simplicity and reliability in linear systems, they often struggle in the presence of nonlinearities and time-varying dynamics, which are common in EV power systems. Hence, augmenting classical control strategies with predictive features offers a promising approach to bridge this performance gap.

Regarding output power, all reported applications operate below 3.5 kW, indicating a focus on low to medium power levels. These levels are typical for on-board systems, portable renewable energy interfaces, and auxiliary EV subsystems rather than high-power applications, such as traction drives or fast-charging stations, which typically require power ratings above 50 kW. The limited power range reflects the current research emphasis on compact, efficient converter designs tailored for distributed and auxiliary functions in electric mobility platforms.

5.4. RQ4: What Are the Reported Applications of Multi-Port dc/dc Power Converter?

The analysis includes a categorization of the reported applications of multi-port DC/DC power converters published between 2006 and 2023. These applications span a broad array of domains, including renewable energy integration, photovoltaic (PV) systems, energy storage, electric transportation, microgrids, aerospace systems, and advanced infrastructure.

Table 6. Categories of applications reported in multi-port dc/dc power converters.

Category	Application	N°	Year	Efficiency	Output Power
Renewable Energy Integration	Integration of renewable energy sources	19	2007–2021	70–90%	90 W–500 W
	Simultaneous Power Management of Multiple Different Renewable Energy Sources	1	2014	86–92%	10 W–100 W
	Renewable Energy and Energy Storage	1	2021	-	-
	Distributed energy sources with grid/load	1	2021	-	1kW
Photovoltaic (PV) Applications	PV system, DC microgrid	1	2022	96.2%	-
	Low Power PV Applications	1	2021	-	-
	Photovoltaic Panels	23	2012–2022	90–99.5%	48 W–3900 W
	Photovoltaic Panels, Battery storage systems	4	2010–2022	-	1kW
	Photovoltaic Panels, Electric Vehicle, Battery storage systems	1	2022	97.7–97.9%	1600 W
	Residential PV-Battery System	1	2020	-	-
	Integration of solar PV, EV fast-charging stations and storage	1	2023	-	-
Energy Storage Systems	Energy Storage Devices	1	2017	-	-
	Energy Storage Devices to DC Microgrid	2	2017–2020	84–96%	800 W
	Battery Energy Storage System	2	2021	93%	250 W
	Hybrid Energy Storage Applications	2	2019–2021	90–99%	1000 W–1500 W
	Hybrid Energy Systems	6	2012–2021	70–98%	120 W
	Hybrid Energy Systems, Storage Systems	2	2016–2021	94–95%	25 W–120 W
Electric Vehicles and Transportation	Electric Vehicle	15	2015–2022	79.3–99.5%	100 W–3300 W
	Plug-in Electric Vehicles with PV and Grid	1	2021	-	-
	DC distribution systems and EV applications	1	2020	-	-
	DC microgrids, EVs and electric aircraft	1	2019	-	-
	Electric aircraft	1	2018	-	-
	DC grid systems	3	2020–2022	68–97%	200 W
Microgrids and DC Grids	DC microgrid systems	5	2018–2022	95–96%	200 W
	DC nanogrids	2	2013–2021	-	250 W
High/Medium Voltage DC	HVDC transmission technology	4	2019–2022	-	2000 W
	Medium Voltage DC (MVDC) Applications	1	2021	-	-
Aerospace and Satellite Systems	CubeSat	1	2022	94–96%	2.25 W–3.1 W
	Satellite Platform Power System	2	2010–2011	86–93%	200 W–800 W
Smart Grids and Infrastructure	Smart grids	1	2020	96%	105 W
	5G centers, EV chargers, audio amplifier	1	2022	94.5–95.4%	500 W
Not reported	Not reported	14	-	-	-

presents the categorized overview, offering insights into application trends, reported efficiency levels, and power capabilities. For key categories such as photovoltaic systems and electric vehicles, two of the most thoroughly studied domains, efficiency metrics, were compiled, where available, to facilitate comparative analysis.

Renewable energy integration is one of the most prominent categories, encompassing 22 reported cases. It includes integration of multiple renewable sources and hybrid energy systems, with efficiencies ranging from 70% to 96.2% and power outputs between 10 W and 3.9 kW. The variation in performance is attributable to differences in converter topologies, control strategies, and operational conditions. However, the predominance of low-power prototypes (500 W) suggests a focus on conceptual validation rather than scalable commercial solutions.

Photovoltaic applications represent the largest application group, with 32 studies in total. Efficiencies in this category are among the highest reported, reaching up to 99.5%, particularly in implementations that integrate energy storage or electric vehicle (EV) interfaces. Power outputs range from 48 W to 3.9 kW, indicating a dominant focus on low- to medium-power systems suitable for residential, microgrid, or auxiliary EV use. Nonetheless, several entries omit full efficiency or power details, revealing inconsistencies in reporting across studies.

Energy storage systems include 15 studies focusing on standalone energy storage as well as hybrid systems that buffer power in PV, grid, or EV contexts. Reported efficiencies span from 84% to 99%, with most implementations operating below 1.5 kW. These systems are primarily intended for load leveling and energy balancing in small-scale networks. Despite the growing importance of energy storage in renewable energy integration, the relative scarcity of high-power applications in this category points to an underdeveloped research area, especially for grid-level or fast-charging systems.

Electric vehicles and transportation, with 22 papers, examine multi-port dc/dc power converters in electric transportation contexts, including EVs, electric aircraft, and vehicular microgrids. Efficiencies reach up to 99.5%, highlighting the potential of these architectures for highly dynamic, efficiency-critical environments. However, all reported implementations operate below 3.3 kW, suggesting a focus on subsystems (e.g., on-board chargers or auxiliary power units) rather than high-power traction or charging infrastructure, which typically exceeds 50 kW. This underscores a research gap in scaling multi-port converters for high-power vehicular applications.

Microgrids and dc grids report seven entries dedicated to grid-level implementations, particularly DC microgrids and nanogrids. Reported efficiencies range from 95% to 96%, reflecting significant variation likely due to load heterogeneity and differing converter control strategies. While multi-port converters offer substantial benefits for flexible energy routing in microgrids, the relatively small number of studies and the constrained power range highlight the need for further research into scalable grid integration strategies.

High and medium voltage dc applications comprise five studies addressing HVDC and MVDC systems despite their strategic importance in modern power transmission. The absence of reported efficiency and power data in these publications may be attributed to three factors. First, the data might be classified as proprietary information and thus not shared. Second, early-stage research may have rendered the findings inconclusive and thus unpublished. Finally, the limitations of the publication process itself may have prevented the dissemination of the data. However, the underrepresentation of high-voltage applications underscores a significant opportunity for further exploration of multi-port converters in large-scale energy infrastructure.

The field called aerospace and satellite systems is composed of three studies that concentrate on CubeSats and satellite platforms. These systems operate at very low power levels (2.25 W to 800 W) yet demand high reliability and efficiency. While the findings demonstrate the adaptability of multi-port converters in space-constrained environments, their limited scope restricts generalization to broader power domains.

Smart grids and advanced infrastructure contains two entries that examine advanced infrastructure use cases, including smart grids and high-performance systems such as 5G data centers. Reported efficiencies exceed 94%, showcasing the potential of multi-port converters for compact, high-efficiency applications in data-intensive and intelligent environments. A notable portion of the literature (14 studies) does not clearly identify the application context.

Photovoltaic (PV) applications have emerged as the most extensively explored domain in multi-port dc/dc power converter research, reflecting global emphasis on solar energy integration. Numerous configurations have been investigated, particularly those that combine Maximum Power Point Tracking (MPPT) with Pulse Width Modulation (PWM) control. A notable example is a novel unidirectional multi-port power converter designed for various PV system configurations, achieving 32 kW, which marks one of the highest power levels reported in the literature. In contrast, smaller-scale applications also demonstrate innovation and precision, such as a 100 W isolated three-port converter integrating a PV module, a bidirectional battery interface, and an isolated output port, an example of compact design with robust energy management capability. Additionally, a 1.6 kW modular bidirectional soft-switching high-gain converter exemplifies advanced topological solutions aimed at improving voltage boosting and conversion efficiency in medium-scale systems. Other studies highlight integration with DC microgrids and PV-battery systems, particularly through Triple Active Bridge (TAB) topologies designed for residential energy management.

Table 7. Highest efficiencies recorded for photovoltaic applications.

ID	Year	Modulation	Control System	Topology	Efficiency
ID565	2016	MPPT	PWM	Non-Isolated	99.5%
ID421	2019	MPPT	PWM	Isolated	98%
ID20	2022	Voltage	PWM	Non-Isolated	97.7%
ID439	2019	PI	PWM	Non-Isolated	96.7%
ID395	2019	PI	QPSM	Isolated	96%

presents the highest efficiency values reported in selected PV applications. The best-performing system (ID565, 2016) achieved 99.5% efficiency using an MPPT-PWM control strategy with a non-isolated topology, emphasizing the potential of non-isolated designs when optimized for minimal conduction and switching losses. Isolated converters, while slightly less efficient (e.g., ID421 at 98% and ID395 at 96%), offer advantages in galvanic isolation and safety-critical applications for grid-connected or high-voltage applications. The control techniques across these high-efficiency implementations include MPPT, PI controllers, and voltage-based control, each paired with PWM or more advanced modulation schemes such as Quasi-Phase Shift Modulation (QPSM). The consistent use of PWM highlights its reliability and effectiveness in PV energy conversion systems.

Table 8. Highest efficiencies recorded for electric vehicle applications.

ID	Year	Control System	Modulation	Topology	Efficiency
ID626	2015	predict-reset PI	-	Non Isolated	99.5%
ID252	2021	PI	PWM	Non Isolated	98.2%
ID20	2022	Voltage	PWM	Non Isolated	97.7%
ID98	2022	P&O-MPPT	PWM	Non Isolated	97.48%
ID129	2022	PI	SPSC	Isolated	97%

presents the five highest reported efficiencies for multi-port dc/dc power converters used in electric vehicle (EV) applications, revealing performance levels comparable to those achieved in photovoltaic systems. EV systems constitute a rapidly evolving domain for converter research, driven by the need for high efficiency, compact integration, and real-time control adaptability.

The highest recorded efficiency, 99.5%, was reported in 2015 (ID626) using a predict-reset PI control strategy applied to a non-isolated topology. This result highlights the potential of optimized predictive and Proportional–Integral (PI) control schemes to minimize dynamic losses and enhance steady-state efficiency in automotive applications, such as onboard chargers and DC-link interfaces. Although the modulation method was unspecified, the performance strongly suggests that non-isolated topologies, characterized by reduced component count and lower conduction losses, offer compelling advantages in vehicular systems.

Subsequent studies, such as those represented by ID252, ID20, and ID98, also employed non-isolated topologies in combination with PWM, PI, voltage, or Perturb and Observe (P&O) MPPT strategies. These configurations achieved efficiency values between 97.48% and 98.2%, reinforcing the suitability of PWM as a dominant and reliable modulation method in EV converters due to its digital compatibility and switching precision.

Only one study (ID129) utilized an isolated topology, reaching 97% efficiency using Single Phase Shift Control (SPSC) paired with a PI controller. Although slightly less efficient than non-isolated designs, isolation is often essential for safety and galvanic separation in specific EV applications, particularly those involving vehicle-to-grid (V2G) or bidirectional charging.

Beyond EV applications, the integration of renewable energy sources—such as wind, hydro, biomass, geothermal, ocean, and photovoltaic systems—has also motivated the design of multi-port dc/dc power converters. These converters support multiple energy pathways and often include bidirectional ports for battery storage integration. In these cases, isolation is mandatory to meet safety and system protection standards. Reported modulation strategies include PWM, as well as hybrid methods such as PWM combined with Phase-Shift Control (PSC) or Peak-Pulse Amplitude Shift (PPAS).

5.5. RQ5: How Many Elements (Passive and Semiconductors) Does the Converter Have?

Semiconductors perform the switching action, while passive elements shape and filter the electrical energy to achieve the desired output and regulate voltage and current changes, protecting the semiconductors. Additionally, in isolated converters, transformers are used to transfer electrical energy between circuits through magnetic coupling.

The quantity of elements that constitute a multi-port dc/dc power converter depends on several factors, including topology, number of ports, power, voltage levels, number of levels, and number of phases.

Table 9. Number of switches registered for automatic controllers in multi-port dc/dc power converters.

N	Controller	Number of Switches	Article ID
1	PI-PSC-PWM	36	ID343
2	PI-PWM	22	ID286
3	PI-PSC-PWM	16	ID519
4	NMPC-PSC	12	ID93
5	MPPT-PSC	12	ID251
6	PI-PSC-PWM	12	ID299
7	PI-QPSM	12	ID395
8	PI-PSC-PWM	12	ID570
9	Current loop-3D - SVM	8	ID177
10	LQR-PWM	8	ID377
11	PID-PWM	7	ID514
12	P&O - MPPT-PWM	5	ID370
13	Predictive-PWM	5	ID614
14	Multi-objective-PWM	4	ID356
15	Fuzzy-PWM	4	ID627
16	PI-Saturation of the i-th duty cycle	4	ID628
17	Time-multiplexed-PWM controllers	3	ID467
18	PI-PFM	3	ID491
19	PI-special PWM switching	3	ID641
20	Multivariable-PFM	2	ID180

presents an extract of the number of switches used by different controllers. This data is obtained from a database compiled in this study, which includes a variety of controllers, starting with the most popular and other combinations. The most commonly reported semiconductors are MOSFETS and IGBTs. Switches are the main devices in power converters (the heart of power converters) because they are the most expensive components and generate the most losses. More detailed data is provided in

Table 10. Number of components used by each converter.

ID	Year	Controller	Modulation	Topology	Eff. %	Application	#C	#L	#SW	#T
ID343	2020	PI	PSC-PWM	Isolated	–	Other	6	0	36 M	2. (3;3)
ID286	2020	PI	PWM	Non Isolated	–	Other	8	3	22 M	0
ID519	2018	PI	PSC-PWM	Isolated	–	Other	4	4	16 I	1. (2;2)
ID299	2020	PI	PSC-PWM	Isolated	96	Other	5	3	12 M	1. (1;2)
ID395	2019	PI	QPSM	Isolated	96	Photovoltaic Panels	15	3	12 M	1. (1;2)
ID251	2021	MPPT	PSC	Isolated	93	Other	3	2	12 I	1. (1;2)
ID201	2021	PI	PWM	Isolated	98.5	Other	5	5	12 I	3
ID93	2022	NMPC	PSC	Isolated	–	Other	3	3	12 M	1. (2;1)
ID236	2021	PI	PSC	Isolated	–	Other	3	3	12M	1. (2;2)
ID334	2020	PI	PWM	Isolated	–	Other	3	3	12 I	1
ID436	2019	PI	PSC	Isolated	–	Other	4		12 I	1.(2;2)
ID535	2017	PI	PSC-PWM	Isolated	–	Electric Vehicle	3	3	12 M	1.(1;2)
ID570	2016	PI	PSC-PWM	Isolated	–	Other	3	2	12 M	1.(2;1)
ID312	2020	PI	PWM	Isolated	95.6	Other	13	4	10 M	1
ID431	2019	MPPT	PWM	Non Isolated	93.7	Photovoltaic Panels	8	2	10 M	0
ID488	2018	PI	PSC	Isolated	–	Other	9	7	10 I	0
ID503	2018	PI	PSC-PWM	Isolated	–	Photovoltaic Panels	3	3	10 M	1
ID592	2016	PI	PWM	Isolated	–	Photovoltaic Panels	6	6	10 I	1.(1;2)
ID518	2018	PS-PWM	PWM	Isolated	–	Other	12	3	9 M	1.(2;2)
ID427	2019	PI	Not Reported	Isolated	98.5	Other	10	5	8 M	1
ID129	2022	PI	SPSC	Isolated	97	Electric Vehicle	0	N	8*N	0
ID177	2021	Current	3D - SVM	Non Isolated	96	Other	4	6	8 M	0
ID135	2022	PI	PSC	Isolated	9–	Electric Vehicle	3	4	8 M	1
ID213	2021	PI	PWM	Isolated	–	Other	7	0	8 M	1.(1;2)
ID377	2019	LQR	PWM	Isolated	–	Other	2	2	8 I	1
ID419	2019	PI	PSC	Isolated	–	Other	2	1	8 I	0
ID478	2018	PI	PSC	Isolated	–	Other	9	5	8 I	1. (2;1)
ID738	2010	MPPT	PWM	Isolated	–	Other	16	3	8 M	1. (2;1)
ID514	2018	PID	PWM	Non Isolated	–	Photovoltaic Panels	5	2	7	0
ID160	2021	PID	PWM	Non Isolated	98	Other	4	4	6 M	2
ID260	2020	PI	Modified PSC	Isolated	96	Electric Vehicle	4	3	6 I	1 (1;2× 1/2)
ID448	2019	PI	PWM and PFM	Isolated	95	Photovoltaic Panels	11	3	6 M	0
ID759	2008	PWM controller	PWM	Isolated	95	Other	7	4	6 M	1.(2;1)
ID739	2010	PID	PWM	Isolated	94	Other	4	2	6 M	1.(1;2)
ID741	2010	MPPT	PWM	Isolated	94	Other	4	2	6 M	1.(1;2)
ID250	2021	PI	PWM	Isolated	–	Other	5 (N-1)	3 (N-1)	6 (N-1) M	1.(1;2× 1/2)
ID370	2019	P	PWM	Non Isolated	–	Other	8	3	6 M	1
ID371	2019	PI	PWM	Non Isolated	–	Other	4	3	6 I	0
ID372	2019	MPPT	PWM	Isolated	–	Other	5	4	6 M	1
ID420	2019	PI	PWM	Isolated	–	Photovoltaic Panels	4	3	6 I	1
ID421	2019	MPPT	PWM	Isolated	98	Photovoltaic Panels	3	3	5 M	1.(1;2)
ID140	2022	PI	PWM	Isolated	96.77	Other	2	2	5 M	0
ID722	2011	MPPT	PWM	Isolated	93	Other	2	2	5 M	1.(1;2)
ID96	2022	P&O-MPPT	PWM	Non Isolated	–	Electric Vehicle	1	2	5 M	0
ID337	2020	PID	PWM	Non Isolated	–	Electric Vehicle	3	1	5 M	0
ID505	2018	PI	Not Reported	Non Isolated	–	Other	1	1	5 M	0
ID614	2015	Predictive	PWM	Non Isolated	–	Electric Vehicle	1	2	5 M	0
ID633	2015	ADP	Not Reported	Non Isolated	–	Electric Vehicle	1	2	5 M	0
ID745	2010	PI	PWM	Isolated	–	Other	3	2	5 M	1.(1;2)
ID626	2015	predict-reset PI	Not Reported	Non Isolated	99.5	Electric Vehicle	4	1	4	0
ID589	2016	MPPT	PSC-PWM	Isolated	97	Other	3	3	4 M	1.(2;2)
ID124	2022	MPPT	PWM	Isolated	95	Electric Vehicle	2	2	4 M	1.(2×1/2; 2×1/2)
ID678	2014	MPPT	PSC	Isolated	93	Other	3	3	4 M	1.(2;2)
ID682	2014	PI	PSC-PWM	Isolated	93	Other	5	4	4 M	1
ID627	2015	Fuzzy	PWM	Isolated	9–	Other	1	3	4 M	1
ID719	2012	PI	PPAS	Isolated	9–	Other	1	2	4 M	1
ID356	2020	Multi-objective	PWM	Non Isolated	82.7	Photovoltaic Panels	2	2	4 M	0
ID197	2021	PI	PWM	Non Isolated	–	Other	2	2	4	0
ID330	2020	PI and ANN	Not Reported	Non Isolated	–	Other	2	1	4 M	0
ID521	2017	PI and SMC	Not Reported	Non Isolated	–	Other	2	2	4 I	0
ID605	2016	PID	PWM+ PSM	Isolated	–	Photovoltaic Panels	2	3	4 M	1
ID628	2015	PI	Saturation of the i-th duty cycle	Non Isolated	–	Photovoltaic Panels	4	1	4 M	0
ID688	2013	PWM	PWM	Non Isolated	–	Other	2	2	4 M	1.(2;2)
ID711	2012	PI	PSC-PWM	Isolated	–	Photovoltaic Panels	4	2	4M. 8 I	1.(2;1)
ID252	2021	PI	PWM	Non Isolated	98.2	Electric Vehicle	3	2	3 M	0
ID98	2022	P&O - MPPT	PWM	Non Isolated	97.48	Electric Vehicle	3	2	3 M	0
ID439	2019	PI	PWM	Non Isolated	96.7	Photovoltaic Panels	3	1	3 M	0
ID179	2021	PI	PWM	Non Isolated	96.6	Other	3	3	3	0
ID441	2019	MPPT	PWM	Non Isolated	95	Other	4	5	3	0
ID265	2020	Fuzzy	PWM	Non Isolated	94.3	Photovoltaic Panels	3	1	3 M	0
ID666	2014	PO MPPT	PWM	Isolated	92	Other	5	4	3 M	1
ID637	2015	PI	PSC-PWM	Isolated	–	Photovoltaic Panels	5	4	3 M	1.(1;2)
ID281	2020	PI	PWM	Non Isolated	–	Photovoltaic Panels	1	2	3 I	0
ID467	2019	Time Mux	PWM controllers	Non Isolated	–	Photovoltaic Panels	3	2	3 M	0
ID491	2018	PI	PFM	Isolated	–	Photovoltaic Panels	5	5	3 M	1
ID641	2015	PI	special PWM switching	Non Isolated	–	Photovoltaic Panels	1	2	3 M	0
ID703	2012	P&O - MPPT	PWM	Isolated	–	Other	4	4	3 M	1
ID765	2007	MPPT	PWM	Isolated	–	Other	3	1	3 M	1.(1;2)
ID565	2016	MPPT	PWM	Non Isolated	99.5	Photovoltaic Panels	4	1	2 M	0
ID109	2022	PID	PWM	Non Isolated	94	Electric Vehicle	2	2	2 I	0
ID88	2022	Fuzzy	SVPWM	Isolated	–	Photovoltaic Panels	2	2	2 I	1.(2× 1/2;1)
ID180	2021	MV	PFM	Non Isolated	–	Other	3	2	2 M	0
ID545	2017	PWM	PWM	Non Isolated	–	Other	2	2	2 M	0
ID692	2013	PI	PWM	Non Isolated	–	Other	N/2	1	N	0
ID363	2020	PI	PWM	Isolated	–	Other	2 (N +1)	N+1	N+1	0

Eff.: Efficiency. # B: Number of batteries used by power converter. # C: Number of capacitors used by power converter. # L: Number of inductors used by power converter. # SW: Number of switches used by power converter. “M” represents MOSFETs and “I” represents IGBTs. # T: Number of transformers, with notation A. (P;Q). A is the number of transformers in the converter. (In cases where P and Q are not specified, the transformer has one primary winding and one secondary winding with a 1:1 ratio.) P: Number of windings at the primary; “×1/2” factor indicates a center tap. Q: Number of windings at the secondary; “×1/2” factor indicates a center tap. N: Represents the number of ports in a modular multi-port dc/dc power converter.

.We highlight two extreme comparisons. A PI controller in an isolated converter with three phases and three bidirectional ports uses 36 MOSFETS [ID343] to deliver between 4 kW and 5 kW. In contrast, a fuzzy controller in a non-isolated converter with four ports, three for different renewable energy sources and one for load, uses only 4 MOSFETS [ID627] and delivers 90 W.

Depending on the application, the most commonly used combinations for battery charging include PI, MPPT, multi-objective, artificial neural networks, and fuzzy time-multiplexed controllers with PWM, PSC, and PFM modulators.

Table 11. Number of articles classified by application and type of automatic controller for multi-port dc/dc power converter.

Application	Number of Articles	Efficiency	Number Switches	Power	Modulation	Controller
Photovoltaic Panels	23	82–99.5%	2/NI–12/I	48 W–32 kW	PWM, QPSM, PWM and PFM, PSC-PWM, SVPWM,	MPPT, Voltage control, PI, Fuzzy, Multi-objective, Time-multiplexed, PID
Integration of renewable energy sources	19	90–97%	3/NI–12/I	90 W–1000 W	PSC-PWM, PWM, phase angle shift (PPAS)	MPPT, PI, PID, Fuzzy, Partial power processing (PPP), LQR, (P&O) MPPT
Electric Vehicle	15	90–98.2%	2/I–12 /I	100 W–3.24 kW	PWM, SPSC, PSC-PWM	PI, Voltage control, P&O—MPPT, MPPT, PID, PPP, Predictive.
DC microgrid systems	5	96–97%	2/NI–22/NI	200 W–250 W	PWM, PFM, PSC	PI, PEA, PI and HESS, MPPT
HVDC transmission technology	4	Not reported	3/ NI	2KVA	PSC-PWM, PWM	PI

presents the various combinations of controllers, modulation techniques, and their applications.

Considering the number of resistors, the PI controller with PSC modulation in a five-port power converter uses seven resistors (ID488), making it the highest in the report. This is followed by the MPPT-PWM configuration (ID421) with five resistors. On the other hand, configurations such as PI-PSC PWM (ID343), PI-PWM (ID286), PI-QPWM (ID395), PI-PSC (ID236), MPPT-PWM (ID431), PI-SPSC (ID129), and P&O MPPT-PWM (ID96) report zero resistors in their structure; precisely among this group, the best efficiency is determined.

In other matters, multi-port dc/dc power converters with the highest number of capacitors in their topology are MPPT-PWM (ID738) with 16 capacitors and PI-QPSM (ID395) with 15 capacitors. On the other hand, the predictive model with PWM (ID505), Fuzzy-PWM (ID614), and PI-PWM/Special PWM (ID281/ID641) reports using only 1 capacitor in their configurations. Additionally, PI-SPSC (ID129) does not include any capacitor.

Furthermore, converters with the major number of inductors are PID-PWM (ID514) with seven devices; PI-PSC (ID488), PI-PWM (ID592), and Current loop-3D-SVM (ID177), all of them with six inductors; on the contrary, MPPT-PWM (ID565), PID-PWM, and PI-PSC (ID 419) report one inductor, and PI-PWM (ID213) and PI-PSC PWM do not report inductors in their configurations.

Isolated converters incorporate high-frequency transformers. For instance, ID343 designed a three-port input–three-port output dc/dc power converter using a PI-PSC-PWM configuration with two transformers, each having three windings on both the primary and secondary sides, representing the highest resource utilization. In contrast, ID88 developed a simpler two-input–output converter with a transformer featuring a center tap on the secondary side. Non-isolated converters, which do not require transformers, have a simpler design and use fewer semiconductors.

The bidirectional power transfer capabilities between output and input ports can increase the number of components, size, and weight of the converter. To mitigate this, some authors consolidate common elements for various ports and replace diodes with switches. Furthermore, the management of high power or voltage levels will necessitate the use of additional filters, and a wide voltage conversion ratio will further compound the size and weight of the converter due to the larger number of components.

6. Discussion and Future Research Trends

The analysis conducted in this study highlights the complexity and diversity inherent in the characterization of multi-port dc/dc power converters. These converters exhibit a broad range of features, such as control strategies, topological configurations, efficiency metrics, application domains, and the number of ports, that collectively define their operational behavior and suitability for various use cases.

A key observation from the literature review is the fragmented manner in which these characteristics are typically reported. Rather than being consolidated in a unified framework, the relevant information is often dispersed throughout different sections of individual publications. This scattered presentation makes it difficult to compare findings across studies, limits the reproducibility of results, and hinders the identification of consistent research patterns or gaps.

To address these challenges, this study proposed and developed a classification scheme that systematically organizes the defining characteristics of multi-port dc/dc power converters. The scheme, derived from a structured data analysis process, is presented in Figure 9. It serves as a visual and conceptual tool that facilitates the recognition of overlooked areas, underexplored converter configurations, and emerging research trends.

By synthesizing and categorizing key converter features into a single scheme, we aim to reduce the time and effort required to assess the state of the art and to aid researchers in designing new studies that address existing knowledge gaps. Furthermore, this classification lays the foundation for a research agenda that encourages systematic exploration of converter technologies, with the potential to enhance innovation and cross-comparison within the field.

A key opportunity lies in the development of systematic literature reviews based on the proposed classification scheme. By incorporating additional levels of granularity within each classification construct, such as detailed subcategories of control strategies, topologies, and applications, it becomes possible to conduct more in-depth analyses of the characteristics that define converter behavior. Beyond evaluating conventional performance metrics such as efficiency, future research should also consider control performance indicators, including dynamic response, stability margins, and robustness to disturbances. These metrics provide a more comprehensive basis for comparing converter performance across different studies.

Another important yet currently underexplored perspective involves examining the materials used in power semiconductor devices. While this study does not explicitly address the influence of semiconductor materials, recent technological advances, such as the use of wide bandgap materials like silicon carbide (SiC) and gallium nitride (GaN), have demonstrated significant improvements in the functionality and performance of multi-port dc/dc power converters. Future work should integrate material characteristics into the classification framework to capture these technological shifts. Furthermore, the classification scheme developed in this work offers a structured foundation for organizing and comparing the existing literature. It can be used to define the scope and criteria of systematic reviews, helping researchers identify underexplored combinations of characteristics and articulate new lines of investigation. In this way, the proposed framework not only supports the identification of knowledge gaps but also contributes to a more cohesive and targeted research agenda for the field.

Focusing on the control strategies employed in multi-port dc/dc power converters, it is evident that classical controllers, such as Proportional Integral (PI) and Proportional–Integral–Derivative (PID) schemes, remain the most widely adopted. Their popularity stems from their simplicity, reliable transient response, and ease of implementation in hardware platforms. These controllers have been consistently reported from early studies to recent works and constitute the majority of control strategies documented in the literature. However, in complex and nonlinear systems, the performance of classical controllers is often suboptimal, prompting the exploration of alternative control approaches. Modern controllers, including model predictive control (MPC), adaptive control, and intelligent control strategies, are increasingly proposed to address these limitations. Despite their potential, these approaches remain underrepresented in current research and collectively account for fewer reported implementations than classical methods. Notably, predictive controllers have been reported in the literature since around 2015, hybrid controllers combining multiple control strategies since 2012, and intelligent controllers, including artificial intelligence (AI)-based techniques such as fuzzy logic, neural networks, and reinforcement learning, have gained visibility only in recent years, primarily since 2020. As shown, this trend underscores the relatively early stage of research into advanced control schemes for multi-port dc/dc power converters.

In addition to adopting modern control schemes, adaptive systems represent a promising direction. These systems adjust controller parameters dynamically based on variations in system behavior, operating conditions, or disturbances. Moreover, AI techniques offer the potential for real-time parameter estimation or tuning, allowing for higher robustness and adaptability in multi-port dc/dc power converter applications. Despite their potential, many of these advanced control methods lack comprehensive experimental validation in realistic scenarios. There is a clear need for further research that evaluates the performance of these strategies using standardized benchmarks and experimental platforms. In particular, future studies should focus on comparing classical and advanced controllers under identical operating conditions, exploring AI methods for real-time adaptation and fault tolerance, and developing frameworks for controller design that include energy efficiency, dynamic performance, and scalability. By addressing these gaps, the research community can move toward more intelligent, adaptive, and high-performance control solutions for modern power conversion systems.

In terms of modulation strategies, pulse-width modulation (PWM) remains the most widely reported technique in the literature. PWM is characterized by high switching frequency, typically implemented using a single carrier signal, and has demonstrated reliable performance, particularly in photovoltaic (PV) applications. Much like classical controllers, PWM has been utilized for decades and continues to be relevant in current designs due to its simplicity and effectiveness. Since 2012, several PWM variants have been proposed with the aim of enhancing output quality, primarily by reducing total harmonic distortion (THD) and minimizing selected harmonics. These include modifications in switching schemes, such as optimized PWM, special PWM switching, and duty cycle saturation techniques. Moreover, alternative modulation methods such as phase-shift modulation (PSM) and frequency-based switching have also been introduced around the same period, offering new control possibilities for multi-port dc/dc power converter topologies. More recently, in 2021, three-dimensional Space Vector Modulation (3D-SVM) was introduced, showing promising results, especially under current control modes. However, this technique has yet to be thoroughly evaluated across a wider range of converter configurations or control strategies. Its limited application highlights a clear research opportunity. A notable gap lies in the integration of advanced modulation techniques, such as PWM combined with methods like variable switching frequency PWM (VSFPWM), predictive phase angle shift (PPAS), pulse frequency modulation (PFM), and predictive saturation control PWM (PSC-PWM), with non-conventional controllers (e.g., model predictive control, fuzzy logic, neural networks). These combinations could significantly enhance system performance in complex and nonlinear environments. Furthermore, model predictive control (MPC) presents a different paradigm by potentially eliminating the need for a separate modulator altogether. MPC can directly generate switching actions based on real-time predictions, thus offering both flexibility and performance advantages in advanced converter applications. While traditional PWM continues to dominate, there is a significant research opportunity in testing and optimizing newer modulation techniques in conjunction with intelligent and adaptive control schemes. Doing so could lead to improved efficiency, reduced harmonic distortion, and better dynamic performance in modern multi-port dc/dc power converter systems.

The main research trends in the field of multi-port dc/dc power converters focus on improving efficiency, ensuring stability, safety, and robustness, achieving higher power density, maintaining well-regulated DC output voltage, and reducing fabrication costs. According to the findings of this study, approximately 58% of reported converters are isolated. These are favored in applications requiring high voltage gain and galvanic isolation, benefiting from soft-switching conditions that reduce switching losses and electromagnetic interference. However, a key research gap lies in the lower efficiency typically observed in isolated converters compared with their non-isolated counterparts. This opens a valuable research opportunity to develop strategies that enhance the efficiency of isolated multi-port dc/dc converters. Future work may involve designing multi-winding, high-frequency transformers with reduced core and copper losses; exploring novel magnetic materials and optimized geometries to minimize losses; and creating accurate equivalent circuit models with measurable parameters to improve system predictability and control. These efforts require a systematic engineering approach, combining materials science, magnetic modeling, and control theory. However, the design of compact, high-efficiency, high-frequency transformers remains technically complex and costly, often leading to bulky converter architectures [37]. In contrast, non-isolated converters, which depend on energy storage elements for voltage regulation, offer simpler design, higher power density, and typically require fewer semiconductors. To enhance voltage gain, current research focuses on switched capacitor topologies, though limited by high current transients; coupled inductors, which introduce voltage spikes and increase system volume; and voltage multiplier circuits, which aim to reduce voltage stress on semiconductors through modular, scalable architectures. These diverse topologies and their associated performance trade-offs highlight the need for a systematic classification scheme to guide future work. This study contributes to that need by proposing a comprehensive classification framework that organizes converter characteristics, such as topology, control strategy, modulation technique, number of ports, and application domain, into a unified structure. There is a notable lack of studies exploring PWM variants in non-isolated topologies, and non-conventional controllers are not widely applied in multi-port dc/dc power converters, highlighting a promising direction for future research.

When evaluating efficiency, MPPT-PWM and PI-PWM controllers are frequently recognized for delivering high performance, particularly in low-power applications. Their simplicity, robustness, and proven reliability contribute to their continued widespread use and documentation in the literature. However, this study reveals that the highest efficiencies reported in recent years have been achieved through advanced control strategies, such as predictive control, nonlinear control, robust control, and artificial intelligence-based approaches. These techniques are often implemented in combination with modern modulation methods, including 3D Space Vector Modulation (3DSVM). There is a growing trend toward testing these advanced control strategies across diverse applications, signaling a shift in research priorities. Despite this, it is important to note that the majority of reported systems operate below 50 kW. This indicates a significant research gap and suggests that further investigation is needed to adapt and validate advanced control and modulation techniques in high-power multi-port dc/dc power converters. Addressing this gap is essential for enabling the next generation of high-efficiency, high-performance power electronic systems suitable for industrial and large-scale renewable energy applications.

As previously discussed, multi-port dc/dc power converters are increasingly being adopted in renewable energy systems due to their flexibility in integrating and managing multiple energy sources and storage elements. This study identifies photovoltaic (PV) applications as the most frequently reported use case, with implementations documented consistently from 2010 to the present. The most commonly used control modulation combinations in these applications include PI + PWM, MPPT + PWM, voltage control + PWM, and PI + phase-shifted PWM (PS-PWM). Less conventional but noteworthy approaches such as PEEA + PWM, PI + quasi-square PWM (QSPM), PI + PWM + pulse frequency modulation (PFM), multi-objective control + PWM, and fuzzy logic + PWM have also been reported, though each appears in only a single publication. This limited exploration points to a clear research gap: there is a significant opportunity to test and validate these emerging control strategies to improve power handling and efficiency in PV systems. A second but less reported application domain is renewable energy resource integration, which has been discussed in the literature since 2007. Similar to PV systems, the predominant control schemes involve MPPT + PSC-PWM, PI + PWM, PID + PWM, and hybrid techniques such as MPPT + PI + PSC-PWM and P&O + MPPT + PWM. Despite the long-standing interest, the research remains heavily reliant on conventional control techniques, underscoring a pressing need for new methods to address increasing complexity and performance demands. Additional emerging application areas include energy storage systems, electric vehicles (EVs), and transportation, where multi-port dc/dc power converters can enable advanced power management across multiple ports. Although various controller configurations have been proposed, efficiency improvement and loss reduction remain unresolved challenges. Notably, robust controllers (e.g., H-infinity or sliding mode) and artificial intelligence-based controllers have not yet been systematically tested in these high-demand contexts, representing another critical direction for future research. Other underexplored but promising fields include microgrids and DC grids, medium/high-voltage DC applications, aerospace and satellite systems, smart grids, and next-generation infrastructure. For instance, the rise of DC distribution in data centers is driven by its higher efficiency and compatibility with modern IT equipment, yet the role of multi-port dc/dc power converters in these systems remains insufficiently examined. Particularly in satellite and CubeSat applications, multi-port dc/dc converters play a vital role in regulating and distributing power among subsystems, optimizing energy harvested from solar panels and stored in batteries. Despite their importance, studies in this area are still scarce, though the growing interest signals an emerging research trend with substantial potential.

The number of switches, which are typically the most expensive components in a converter, is directly related to the number of ports in the system. Passive elements serve primarily as signal filters and auxiliary components, providing protection to semiconductors against excessive current and voltage transients, including high rates of $dv/dt$ and $di/dt$. Advanced controllers offer the potential to simplify the design of multi-port dc/dc power converters by improving system dynamics and potentially reducing the number of required devices. This benefit has driven increasing research interest in the application of non-classical control strategies to dc/dc converter design. However, a significant limitation is that advanced control algorithms typically demand greater computational resources, which can lead to increased system costs. Therefore, a key research challenge is to develop advanced controllers with lower computational requirements. This can be addressed either through algorithmic optimization or by leveraging ongoing technological advancements to create more powerful and cost-effective hardware platforms with enhanced capabilities.

Finally, a trend in power semiconductor development is the exploration of new materials, particularly silicon carbide (SiC) and gallium nitride (GaN) power switches, which provide high speeds and low switching losses. Another approach involves the utilization of soft-switching techniques. However, challenges persist in ensuring electromagnetic interference (EMI) resistance, reducing voltage and current peaks during switching, and minimizing fabrication costs.

7. Threats to Validity

The study under development is subject to various conditions that pose threats to its validity. This section identifies and addresses each of these potential threats to validity detected during the study. Each stage was carefully developed by dismissing them.

7.1. Construct Validity

In establishing the construct validity of a systematic mapping study, the absence of consensus in distinguishing nomenclature between the controller and modulation technique necessitated a thorough review and classification of existing literature. For the first stage of defining the database search strategy, the most suitable and complete database for our research, Scopus, was selected. The work in [38] states that Scopus is better than Web of Science for identifying computer science publications; it claims to be the largest abstract and citation source of peer-reviewed literature, including articles from ACM, IEEE, and Springer (the most relevant catalogs for computer science).

7.2. Internal Validity

For facing threats to validity related to the search string, it was identified as shown in Section 4.2, and then it was validated from the list of terms that express the aim of each section. For the term "dc/dc converter," the 2017 IEEE Thesaurus Version 1.0 determines a classification. Definitions of terms for describing reusable elements were obtained from a bibliographic revision of the literature on the area. The results were iteratively validated against two test sets of 10 papers proposed by an experienced power electronics researcher. After each validation, the search string was improved. The final search string demonstrated the major quantity of papers recommended appeared after a search run.

Criteria definition for inclusion and exclusion and the procedure for applying them constitute another threat to validity. In this case, the research team (one screener and one supervisor) met, updated, and agreed on the criteria and the procedure to follow. In addition, the criteria to be used in the manual procedure were validated during the pilot phase of the inclusion/exclusion procedure.

To address threats to validity, the manual screening procedure was carried out by one screener under the supervision of a supervisor. From an initial pool of 753 papers, a pilot stage was first conducted to standardize the knowledge of the screener. To reduce human bias, the inter-coder reliability and kappa coefficient were calculated, with a success threshold set at 0.8, a value accepted in most studies. During this pilot phase, two to three iterations were performed using test sets of 35, 10, and 7 papers (5% of the total) for title screening, with fewer papers used for abstract and full-text screening. After each iteration, the screener and supervisor met to resolve any divergences.

In the main screening stage, the screener worked on a larger set of papers. When the kappa coefficient fell below the threshold, the team discussed the divergences. In the second iteration, the “unclear” papers from the first iteration were re-screened, this time including the conclusions section for a deeper understanding. During the third iteration, both the screener and supervisor jointly decided whether to include or exclude each paper, thereby reducing bias.

Regarding classifications used for coding, well-accepted proposals were employed to mitigate threats to validity. For research types, we used the classification proposed by [39]. The definitions of reusable elements for coding were based on works by [1,2,40].

7.3. External Validity

The primary threat to the validity of the coding procedure is human bias. This can be mitigated by involving supervisors in the screening process and creating a formal codebook. If any doubts arose, the paper was marked as unclear and discussed with the supervisor.

Other threats to the validity of the study were identified. One issue is that the classification scheme lacks depth, as there are no formal studies to support it. This results in a very general study, making it difficult to draw specific conclusions. Additionally, published control systems often do not clearly distinguish between the controller and the modulation technique, and there is no consistent terminology used by authors to refer to the topology. To mitigate this threat, theoretical foundations were used to justify the classifications, allowing each study to be accurately categorized within them.

8. Conclusions

This study presents a research agenda for multi-port dc/dc power converters developed through a systematic mapping study. It investigates controllers, topologies, modulation techniques, applications, efficiency, and component quantities. This study compiles articles published in Scopus from 2005 to 2023. Previous similar studies focus narrowly on specific applications and types of power converters. They do not encompass all the characteristics that this study defines for power converters.

The research agenda emphasizes the challenges faced by multi-port dc/dc power converters in terms of their design based on defined characteristics. It aims to encourage researchers to explore new configurations and novel methodologies in this field, facilitating the identification of optimal efficiencies and applications for potential adaptation into other contexts. Additionally, it aids in identifying configurations that may require modification to enhance performance.

An interesting improvement to the current review involves considering additional features of the power converter, such as power, power factor, dimensions, and voltage levels. Artificial Neural Networks and fuzzy logic are expected to provide effective solutions for simplifying controllers and reducing the number of components. Additionally, it would be beneficial to identify whether the proposed converter is implemented or simulated.

Finally, understanding the latest advancements in control systems for multi-port dc/dc power converters can help propose alternatives that enhance performance and improve energy efficiency.