Energy Management Technologies for All-Electric Ships: A Comprehensive Review for Sustainable Maritime Transport

Abstract

To systematically review the research progress, methodological frameworks, and application characteristics of energy management technologies for All-Electric Ships (AES), this review provides a comprehensive and critical survey of studies published over the past two decades, following the technical trajectory of multi-energy coupling–multi-objective optimization–engineering-oriented operation. Based on a structured analysis of representative literature, the review first elucidates the overall architecture and operational characteristics of AES energy systems from a system-level perspective, highlighting their core advantages as “mobile microgrids” in terms of multi-energy coordination and dispatch flexibility. On this basis, a structured classification framework for energy management strategies is established, and the theoretical foundations, applicable scenarios, and engineering feasibility of rule-based, optimization-based, uncertainty-aware, and intelligent/data-driven approaches are comparatively reviewed and discussed. Furthermore, focusing on key research themes—including multi-energy system optimization, ship–port–microgrid coordinated operation, battery safety and lifetime-oriented management, and real-time energy management strategies—the review synthesizes the main findings and engineering validation progress reported in recent studies. The analysis indicates that, with the integration of fuel cells, renewable energy sources, and Hybrid Energy Storage Systems (HESS), energy management for AES has evolved from a single power allocation problem into a system-level optimization challenge involving multiple time scales, multiple objectives, and diverse sources of uncertainty. Optimization-based and Model Predictive Control (MPC) methods have shown promising performance in many simulation and pilot-scale studies for improving energy efficiency and emission performance, while robust optimization and data-driven approaches offer useful support for enhancing operational resilience, prediction capability, and decision quality under complex and uncertain conditions. These advances collectively contribute to the environmental, economic, and operational sustainability of maritime transport by reducing greenhouse gas emissions, extending equipment lifetime, and enabling efficient integration of renewable energy sources. At the same time, the current literature still reveals important limitations related to model fidelity, data availability, validation maturity, and the gap between methodological sophistication and practical deployment. Overall, an increasingly structured but still evolving research framework has emerged in this field. Future research should further strengthen ship–port–microgrid coordinated energy management frameworks, develop system-level optimization methods that integrate safety constraints and uncertainty, and advance intelligent Energy Management Systems (EMS) oriented toward sustainable zero-carbon shipping objectives.

1. Introduction

With the growing global demand for green, low-carbon, and intelligent development of the shipping industry, issues such as low energy efficiency, excessive greenhouse gas emissions, and high operational costs of conventional vessels have become increasingly prominent, posing major challenges to the sustainable development of maritime transportation. To effectively address these challenges a sustainability perspective, the All-Electric Ships (AES) concept has emerged. AES represents a revolutionary solution in the maritime domain aimed at achieving efficient, low-carbon, and intelligent navigation. Its core characteristic lies in the adoption of an integrated power system to supply both propulsion and shipboard service loads, forming a typical “mobile microgrid.” In recent years, with the rapid advancement of power electronics, large-capacity energy storage, and artificial intelligence technologies, energy management and optimization of AES become a critical technological frontier for promoting green shipping and supporting national maritime development strategies.

In recent years, both domestic and international scholars have conducted extensive research in the field of energy management for AES. In earlier reviews, Hou et al. (2018) systematically compared two technological approaches, battery-supercapacitor and battery-flywheel systems, and thoroughly analyzed the performance boundaries of various energy management strategies [1]. That same year, the team further conducted an in-depth performance evaluation of the battery-flywheel Hybrid Energy Storage System (HESS), developing a real-time Model Predictive Control (MPC) strategy [2]. As research advanced, scholars began to focus on addressing the uncertainties in system operations. In 2020, Fang et al. were the first to introduce a two-stage robust optimization method into AES energy management, effectively addressing uncertainties in navigation speed caused by wind and waves, as well as the risk of port arrival delays [3]. In the same year, Wen et al. integrated Photovoltaic (PV) generation into the optimization framework and used deep learning techniques for PV power forecasting, marking the beginning of a new research direction combining “renewable energy + data-driven approaches + navigation scheduling [4].” In 2021, Kim et al. provided a system engineering design guide, and Zhao et al. (2022) advanced the field with a distributed robust model, further promoting the deep integration of “source-network-load-storage-navigation [5,6].” Recent studies have shown a diversification in research approaches. In 2022, Hein, Kyaw established a new benchmark for multi-objective optimization through the integration of LSTM and NSGA-III [7]. Important progress in new power systems has been made, with Cao et al. (2023) [8] and Jung et al. (2023) [9] exploring the integration of hydrogen fuel cells and deep reinforcement learning. On the engineering practice front, Chen et al. (2023) validated the engineering feasibility of adaptive MPC through hardware-in-the-loop testing [10], and Pang et al. (2024) combined Extended Kalman Filtering (EKF) with MPC to achieve real-time optimization for LNG-battery hybrid power systems [11]. In the same year, Evaggelia et al. demonstrated the effectiveness of fuzzy logic for near-optimal control in fuel cell-powered ships [12]. Additionally, the thermal safety design by Liu et al. (2024) and the lifecycle cost approach by Li et al. (2025) collectively signaled a profound shift in the field from performance optimization to engineering reliability and lifecycle management [13,14].

Despite the substantial progress achieved in both technical research and engineering applications of energy management for AES, numerous challenges remain. The sustainable operation of fuel cell-based all-electric ships depends critically on the availability of green fuel. Reversible solid oxide cells have emerged as a promising technology that can address this challenge by operating alternately in electrolysis mode to generate hydrogen using intermittent renewable energy (e.g., offshore wind, solar) and in fuel cell mode to convert the stored hydrogen back into electricity [15,16]. Such systems offer high round-trip efficiency and the potential to reduce reliance on large battery banks, making them particularly attractive for shipboard applications where long-term energy storage and fuel flexibility are essential. However, the integration of Reversible solid oxide cells technology into shipboard power systems remains underexplored, representing a critical research gap. For instance, the dynamic coupling mechanisms between shipboard power systems and propulsion loads have not yet been fully understood, and the capacity sizing and power allocation of HESS still require further optimization [17]. In addition, existing review studies have largely focused on categorizing and summarizing various energy management methods, with an emphasis on textual descriptions, while lacking visualized approaches that could provide an intuitive and comprehensive understanding of the research landscape. To clarify the research gap addressed by this review,

Table 1. Summary of Representative Review Articles on Energy Systems.

Title	Year	Topic	Core Content	Coverage	Difference
A review of recent advances on Hybrid Energy Storage System for solar photovoltaics power generation [18]	2022	HESS for Solar PV	Capacity sizing, power converter topologies, energy management strategies for HESS in PV power generation	Focuses on HESS for PV; limited to grid-connected/off-grid PV applications	Does not address all-electric ships or multi-energy coordination with fuel cells
Component and system levels limitations in power-hydrogen systems: Analytical review [19]	2024	Hydrogen Energy Systems (HES)	Component-level limitations (production, storage, transportation) and system-level integration challenges with electric power systems	Comprehensive on hydrogen technology; focuses on hydrogen as energy carrier	Lacks integration with shipboard power systems and energy management strategies specific to AES
Review of optimal design and enhanced Hybrid Energy Systems using Energy Management Strategies [20]	2025	HES Design and EMS	Optimal system design, modeling approaches, energy management strategies (rule-based, MPC, AI-based) for hybrid renewable systems	Broad coverage of hybrid systems; includes both standalone and grid-connected applications	Does not specifically address all-electric ships; limited focus on ship-port-microgrid coordination
Systematic review of battery electric ship safety: risk factors, assessment methods, and preventive measures [21]	2025	Battery Electric Ship (BES) Safety	Risk factors (seafarer, ship, environment, management), risk assessment methods, preventive measures for BES safety	Systematic review of BES safety; covers risk identification and mitigation	Focuses on safety rather than energy management optimization; lacks EMS strategy classification

summarizes representative review articles published in the past five years on shipboard energy management and related hybrid energy systems. As shown, existing reviews tend to focus on specific aspects such as HESS for photovoltaic applications [18], hydrogen energy systems at component and system levels [19], optimal design and energy management strategies for hybrid systems [20], or battery electric ship safety [21]. However, these studies either concentrate on a single technology domain, lack a systematic categorization of energy management strategies for all-electric ships, or do not comprehensively cover the full spectrum from system architecture to engineering-oriented real-time operation. In contrast, this review provides a systematic analysis spanning multi-energy coupling, multi-objective optimization, and engineering-oriented operation, with a particular emphasis on cross-category comparisons and the practical challenges of onboard deployment.

This review focuses specifically on energy management technologies for AES, including hybrid electric ships with significant electric propulsion components. It covers onboard EMS and coordinated ship-port-microgrid operation, while excluding studies on conventional mechanical propulsion or generic energy systems lacking marine-specific context. The scope is confined to literature addressing maritime-specific challenges, such as sea-state disturbances, load fluctuations, and the integration of renewable energy and energy storage in shipboard power systems. Accordingly, this review aims to critically examine research progress in AES energy management, encompassing theoretical methodologies, technological frameworks, engineering applications, existing challenges, and future directions. By providing an in-depth understanding of the critical role of energy management technologies in AES development, it seeks to offer valuable theoretical references for technological innovation and engineering practice in this field.

The remainder of this review is organized as Figure 1 follows. Section 2 describes the research methods and data collection process. Section 3 presents the bibliometric analysis and results. Section 4 reviews and analyzes the current hot research areas in the field of energy management for AES. Section 5 proposes the existing challenges and future opportunities, in combination with the current technological status. Finally, Section 6 provides the conclusions of this review.

2. Electric Ship Energy Systems and Operational Characteristics

2.1. Overall Architecture of Electric Ship Energy Systems

The energy system of an electric ship typically comprises four hierarchical layers: energy supply, energy conversion and storage, energy-consuming loads, and energy management and control. Together, they form an integrated architecture of “source–grid–load–control”. In contrast to traditional engine rooms relying on diesel engines as the sole primary power source, electric and hybrid electric ships incorporate diverse energy sources on the supply side, such as shore power connections, fuel cells, and photovoltaics. These sources supply power to propulsion and service loads via a unified electrical bus, enabling physical decoupling of the propulsion and power supply processes while allowing for coordinated optimization in scheduling.

Structurally, the energy supply layer includes shore power/charging facilities, diesel generator sets or fuel cell systems, and renewable energy installations. The energy conversion and storage layer establish AC/DC buses through rectifiers, converters, and transformers, integrating devices like lithium-ion batteries, supercapacitors, and hydrogen storage systems. These are used to mitigate load fluctuations and achieve load shifting across different time scales. The energy-consuming load layer mainly consists of propulsion motors, thrusters, electric steering gears, as well as hotel loads like air conditioning, lighting, and refrigeration. The high-reliability loads may be connected via independent feeders or redundant busbars.

Spanning all these layers is the energy management and control system, generally composed of the onboard EMS, Power Management System (PMS), and various local controllers. On one hand, the EMS allocates the output power and charge/discharge schedules of each energy unit based on voyage plans, sea conditions, and operational states. On the other hand, the PMS is responsible for maintaining bus voltage and frequency stability and ensuring power supply to critical loads under fault or islanded operating conditions. Figure 2 provides a schematic diagram of the overall architecture of electric ship energy system, establishing the systemic structural context for the discussion of different energy management strategies in Section 3.

2.2. Characteristics of Main Energy-Conversion and Energy-Storage Units

Electric-ship energy systems are commonly conceptualized as a coordinated architecture comprising generation units, energy storage, and external or renewable energy supplementation. Across the reviewed studies, configurations featuring fuel cells coupled with battery-based energy storage recur frequently, often further integrated with shore power and PV sources. Within robust co-scheduling frameworks, fuel cells and batteries are treated not only as power-supplying components but also as assets whose degradation/aging costs are explicitly represented, such that energy management balances short-term operating economics against long-term reliability and lifetime-related losses. In addition to hydrogen, ammonia has recently attracted significant research attention as a promising maritime fuel. Green hydrogen produced via electrolysis can be further utilized for ammonia synthesis, offering a carbon-free energy carrier that is easier to store and transport than hydrogen. The integration of ammonia as a fuel or hydrogen carrier into shipboard multi-energy systems represents an emerging direction for future research. In shipboard multi-energy power systems, coordinated operation among CHP, hydrogen fuel cells, batteries, hydrogen storage, LNG, and PV is typically modeled at the system level, where PV variability is identified as a key disturbance requiring robust representations of uncertainty. Strategy effectiveness is then reported through percentage-based cost changes, including cases where total operating cost decreases by 4.92% while the worst-case cost increases by only 0.38%. Recent studies have also explored the integration of artificial intelligence in renewable energy and storage systems to enhance performance and decision-making [22].

In terms of thematic emphasis, this study adopts a structured review approach that combines bibliometric mapping with qualitative critical synthesis. Literature was retrieved from the Web of Science Core Collection using the search strategy summarized in

Table 2. Literature Search Parameters.

Parameter	Content
Database	Web of Science Core Collection—SCI-EXPANDED and SSCI
Time range	2002–2025
Document type	journal article OR review article English
Language
Search formula	Topics = (“new energy ship” OR “electric ship” OR “All-Electric Ship” OR “AES” OR “Hybrid Electric Ship” OR “HES”) Topics = (“Energy management” OR “Energy management strategy” OR “EMS” OR “Energy storage system” OR “ESS” OR “Hybrid Energy Storage System” OR “HESS” OR “Battery energy storage system” OR “BESS” OR “Battery Management System” OR “BMS”)

. The initial search yielded 520 records. The records were then screened in two stages: title/abstract screening and full-text assessment. Studies were retained if they directly addressed energy management, power allocation, hybrid energy storage coordination, uncertainty-aware scheduling, or data-driven operational optimization in all-electric or hybrid-electric ships. Records were excluded if they focused primarily on non-maritime systems, ship design without substantive energy-management content, logistics/transport management without shipboard energy relevance, duplicate records, or studies lacking sufficient methodological information for classification. After screening, 363 papers were retained, including 346 research articles and 17 review papers. The final corpus was analyzed through bibliometric keyword statistics and thematic critical comparison across rule-based, optimization-based, uncertainty-aware, and intelligent/data-driven strategies.

Table 3. Distribution of High-Frequency Keywords (Energy System).

Keyword	Share (%)
Optimization	13.99%
Energy Management	8.11%
Batteries	6.68%
Storage	5.41%
Microgrids	5.09%
Power Systems	5.09%
All-Electric Ships (AES)	4.93%
Fuel Cell	4.13%
Renewable Energy	4.45%

presents the distribution of high-frequency keywords in the field of AES energy systems, where “Optimization” accounts for the highest proportion (13.99%), followed by “Energy Management” (8.11%) and “Batteries” (6.68%). The high frequency of these keywords directly reflects the core research focus of the current field, with optimization techniques and energy management as the core, and batteries and other energy storage devices as important research carriers.

Figure 3 and Figure 4 present the keyword co-occurrence network and annual trend evolution network, with a minimum occurrence threshold set to eight, comprising 60 nodes, 1043 links, and four clusters. Keywords are represented as nodes, and the connections between nodes reveal their associations and co-occurrence frequencies.

From Figure 3, the research hotspots in this field can be divided into four main clusters. For example, the blue cluster focuses on the coordination and optimization of multi-energy systems, covering topics such as solar energy and wind energy, combined with keywords like “Hybrid Energy Systems”. This indicates that research emphasis lies in improving energy utilization efficiency. The red cluster primarily revolves around electric ship propulsion and EMS, with core content involving energy allocation, propulsion efficiency, and system design. Among these, keywords such as “Energy Management” reflect the current research focus on improving power system efficiency, reducing energy consumption, and enhancing control strategies. The green cluster emphasizes operational and control issues under renewable energy integration, with keywords including microgrids, state of charge, operating costs, and uncertainty, reflecting research attention to system stability and economic performance under renewable energy fluctuations. The yellow cluster is closely related to the development of all-electric and hybrid vessels, involving fuel cells, simulation analysis, emissions, and technical strategies, indicating that structural innovation and performance optimization of power systems continue to receive sustained attention in the context of green shipping. In the overall network, the clusters are closely connected through methods such as “Optimization” and “Energy Management”, highlighting the central and cross-cutting role of optimization techniques and energy management in ship energy system research.

Finally, from the perspective of keyword annual evolution trends, research hotspots can be observed to have gradually evolved between 2020 and 2023. Dark-colored nodes such as “Power-Systems” are concentrated in system infrastructure and renewable energy integration, indicating that early research primarily focused on energy system modeling and generation-side scheduling. As time progressed, keywords transitioning to green and yellow, such as “Design”, “Optimization”, “Management”, “Energy Management”, and “Microgrids”, became core themes in 2021–2022, reflecting that optimization methods, energy management strategies, and microgrid structures gradually became mainstream research topics. Entering 2022–2023, bright yellow nodes such as “State of Charge”, “Algorithm”, “Uncertainty”, “Costs”, and “Storage” significantly increased, indicating that research hotspots have further extended toward uncertainty analysis in electric ships and HESS, state of charge management, operating cost optimization, and energy storage scheduling. Moreover, keywords such as “All-Electric Ships”, “Hybrid Energy Systems”, and “Fuel-Cell” have also shown an upward trend in recent years, indicating that electric ships, fuel cell propulsion, and hybrid energy storage structures are becoming new growth areas.

2.3. Operating Conditions and Energy-Demand Characteristics of Ships

Ship operating conditions fundamentally shape the strong time-varying and uncertain nature of onboard energy demand. Sea-state disturbances, renewable output variability, and load fluctuations jointly alter the power requirements of propulsion and service loads, which in turn can translate into speed loss, delay risks, and operating-cost volatility. Accordingly, energy management formulations frequently incorporate uncertainties in wave speed and direction induced by wind and currents, using robust decision frameworks to maintain feasibility and operational stability under different disturbance realizations [23]. Under more comprehensive disturbance settings, robust approaches further integrate joint uncertainties stemming from wind, waves, and PV, while simultaneously embedding operational objectives and constraints such as on-time arrival and greenhouse-gas limits into co-scheduling decisions.

From an operational perspective, the reviewed literature also emphasizes differentiated energy-consumption patterns between at-sea and at-berth scenarios, and highlights the importance of meteorological variables for at-sea prediction. When sufficiently rich weather information is included, reported at-sea forecasting accuracy improvements range from 4.49% to 16.46%. Beyond prediction, data-driven work increasingly advances toward a prediction–optimization loop: interpretable forecasting is used to identify adjustable operating parameters and to guide voyage-process optimization, with reported energy-efficiency improvements reaching 6.41% and 7.05%. In parallel, monitoring capabilities such as short-horizon roll/pitch prediction and abnormal-behavior identification are also leveraged to support safe operation and adaptive strategy adjustment under varying sea conditions.

3. Classification and Analysis of Energy Management Strategies for All-Electric Ships

3.1. Rule-Based Energy Management Strategies

In the research of EMS for electric and hybrid electric ships, rule-based strategies remain the most widely applied category in engineering practice. On one hand, these strategies are typically built upon the shipboard PMS, controlling the logical switching of components like generator sets, fuel cells, and energy storage units by setting several start-stop rules, state thresholds, and operational modes. On the other hand, rule-based strategies feature clear structures and intuitive parameter meanings, making them easy for crew members to understand and adjust, and often better aligned with classification society requirements for safety and interpretability. A recent review of EMS for hybrid ships points out that rule-based and hierarchical control methods dominate in industrial applications, while optimization-based and intelligent methods are more commonly in the testing and demonstration stages, reflecting, to some extent, the engineering “mainstream status” of rule-based strategies [24].

From a control structure perspective, rule-based strategies can be broadly categorized into three types: (1) Threshold/state machine control, triggered by core variables like State of Charge (SOC) or power [25,26]; (2) Experience-based rule control incorporating fuzzy logic and expert knowledge; and [27,28]; (3) Hierarchical rule strategies, often coupled with the existing PMS, which combine offline optimization or vessel operation experience [24,26]. Based on a review of representative literature, the typical approaches, advantages and disadvantages, and applicable scenarios for these three strategy types are summarized and compared in

Table 4. Comparative Overview of Rule-Based Energy Management Strategies for Electric and Hybrid-Electric Ships.

Representative Method	Vessel/Application	Validation Level	Computational Burden	Main Strength	Main Limitation
SOC/Power Threshold EMS	Small passenger ships, inland vessels, hybrid retrofit projects	Simulation/Case-based engineering studies	Low	Simple structure, fast response, easy integration with existing PMS, reduced start-stop frequency	Thresholds are often empirical, with limited optimization capability and weak adaptability to complex operating conditions
Fuzzy-Logic EMS	Passenger ferries and tourist ships with PV or fuel cells	Simulation/Limited experimental studies	Low–Medium	Flexible representation of operator knowledge, low model dependence, mitigates hard switching	Requires extensive tuning, rule design is experience-dependent, and generalization is limited
Hierarchical rule-based EMS	Medium and large commercial ships with strict safety and economy requirements	Simulation/Engineering-oriented case studies	Medium	Balances economy and dynamic response, highly interpretable, suitable for layered shipboard control architecture	More complex control structure and calibration burden than simple threshold-based strategies

.

Basic SOC/power threshold strategies determine logic through preset fixed thresholds. Their advantages lie in clear logic, high real-time performance, and low dependence on system models, making them suitable for small/medium-sized vessels or retrofit projects. However, threshold setting relies on experience, and they have limitations in multi-objective trade-offs and adaptability to varying conditions. To overcome the rigidity of “hard thresholds,” fuzzy logic rule strategies were introduced. By using membership functions and fuzzy inference to translate expert knowledge into flexible control laws, they enhance multi-variable coordination capabilities. Yet, the design of their rule bases and functions still requires extensive debugging, limiting their scalability. For increasingly complex hybrid power systems, hierarchical rule strategies have been further developed. Their upper layer makes power allocation decisions at the mission/voyage scale, while the lower layer, combined with PMS, achieves fast dynamic tracking, thereby balancing long-term economy and short-term dynamic performance. These three categories of rule strategies represent an evolution from solidified experience to intelligent flexibility, jointly providing practical solutions for ship scenarios with varying complexity and real-time requirements.

In summary, rule-based energy management strategies still hold an irreplaceable engineering position in current electric/hybrid ships. Their main advantages include: (1) Simple structure, low implementation cost, and compatibility with existing PMS architectures; (2) Transparent control logic, easy for crew to understand and intervene with, facilitating classification society certification; (3) Lower requirements for system parameters and predictive information, suitable for vessels with limited sensing and communication capabilities.

From an analytical perspective, rule-based EMS is primarily valued for its deployability rather than its optimization capability. These methods work best when control objectives are stable, system behavior is well understood, and computational overhead is a priority. However, their performance depends heavily on manually designed thresholds and rules. This limits adaptability in scenarios involving multiple objectives, rapid condition changes, or significant uncertainty. Therefore, the current evidence supports rule-based strategies as robust engineering solutions for practical implementation, but not as the most flexible framework for next-generation AES coordination problems.

However, when facing more complex objectives and constraints (such as multi-energy coordination, emission constraints, component degradation costs, etc.), rule-based strategies reveal shortcomings in lacking systematicity and adaptability. In scenarios with high uncertainty in sea conditions and loads and frequent changes in operating conditions, thresholds set solely based on experience struggle to guarantee global optimal performance. These issues directly prompted the subsequent research on optimization and MPC-based energy management, providing the logical starting point for the discussion in Section 3.2.

3.2. Optimization-Based Energy Management Strategies

Compared to rule-based strategies, optimization-based energy management methods emphasize a systems perspective, explicitly incorporating the operational objectives and constraints of the ship’s propulsion and power systems into mathematical models, and solving for the “optimal” power allocation scheme via optimization algorithms. Typical objectives include minimizing fuel (or hydrogen) consumption, minimizing operating costs, minimizing greenhouse gas emissions, or weighted combinations of multiple objectives. Main constraints include power and ramp limits for generators and fuel cells, SOC ranges for energy storage [29], safety reserve requirements, and speed/voyage constraints.

From a solution perspective, optimization-based EMS can be broadly categorized into three types: (1) Finding global or near-global optimal scheduling via methods like dynamic programming, bi-level optimization, or mixed-integer programming, given predetermined voyage profiles and parameters [30,31]; (2) Based on the Equivalent Consumption Minimization Strategy (ECMS), approximating the “instantaneous optimal” control law at each time step [32,33]; (3) Utilizing MPC [34,35] or Nonlinear MPC (NMPC) to perform rolling optimization of control sequences within a finite prediction horizon [36,37,38]. This section will review representative studies from the three aspects of bi-level optimization, ECMS, and MPC, with a comparison provided in

Table 5. Comparative Overview of Optimization-Based Energy Management Strategies for Electric and Hybrid-Electric Ships.

Representative Method	Vessel/Application	Validation Level	Computational Burden	Main Strength	Main Limitation
Bi-level/Multi-objective optimization	Configuration design and offline scheduling of novel hybrid ships	Simulation/Design-stage optimization studies	High	Supports lifecycle and voyage-scale trade-off analysis for cost, emissions, and configuration design	Computationally intensive and generally unsuitable for direct real-time control
ECMS/Adaptive ECMS	Operational EMS for fuel-cell or diesel–battery–supercapacitor ships	Simulation/Controller-oriented case studies	Low–medium	Low computational cost and good real-time applicability for instantaneous power split	Performance is sensitive to equivalence-factor calibration and does not guarantee global optimality
MPC/NMPC	Online energy management of medium-to-large hybrid ships	Simulation/Hardware-in-the-loop/Limited engineering validation	Medium–high	Systematic multi-constraint handling, natural feedback capability, suitable for rolling-horizon coordination	Sensitive to model fidelity and prediction quality, with relatively high computational burden
Multi-layer MPC + data-driven coordination	Ferries and special-purpose ships with large load fluctuations	Simulation/Integrated control case studies	High	Supports long-short-term coordination and improved robustness to stochastic load variation	High implementation complexity and stronger hardware/Software integration requirements

.

Taken together, the review shows that optimization-based EMS offers clear analytical advantages in multi-objective coordination, explicit constraint handling, and system-level scheduling. However, these advantages are typically realized under assumptions of adequate model fidelity, parameter calibration, and computational tractability. In addition, a substantial proportion of the reported evidence is still based on simulation and controlled validation environments rather than long-duration ship-board deployment. As a result, optimization-based methods are currently supported more strongly as a rigorous decision-making framework than as a universally mature engineering solution.

Compared to rule-based strategies, optimization-based energy management methods offer the following significant advantages: (1) Explicit multi-objective modeling: Fuel/hydrogen consumption, emissions, equipment degradation, passenger comfort, etc., can be incorporated into a unified cost function or multi-objective framework, facilitating transparent trade-offs between different performance metrics; (2) Strong constraint handling capability: MPC/NMPC and mixed-integer programming can naturally handle various constraints like generator power limits, SOC constraints, speed/voyage requirements, and provide interpretable “binding” information for these constraints; (3) Quantifiable and verifiable performance: The vast majority of optimization-based EMS literature provides clear comparisons with rule-based or empirical strategies, quantifying fuel savings or cost reduction percentages (e.g., 2–6% fuel saving, ~5% GHG emission reduction), offering a quantitative basis for engineering decisions.

At the same time, their limitations should not be overlooked: (1) Strong dependency on system models, load, and sea condition predictions. Performance can degrade significantly in the presence of large model errors or unpredictable disturbances; (2) Computational complexity escalates rapidly with system size and prediction horizon, especially for optimization problems involving mixed-integer variables or hierarchical structures, posing challenges for real-time implementation; (3) In engineering applications, acceptance by ship owners and crews for “black-box” optimization strategies remains limited. Finding the balance between interpretability, safety, and optimality is still an important research direction.

It is precisely these limitations that have driven the development of subsequent methods discussed in Section 3.3, such as stochastic optimization, robust optimization, and distributionally robust optimization: by explicitly characterizing uncertainties in loads, sea conditions, and renewable generation, they attempt to find a more appropriate compromise between “optimality” and “robustness.”

3.3. Energy Management Methods Under Uncertainty

Within the classic “four-way taxonomy” of energy management strategies, this subsection focuses on approaches that explicitly incorporate uncertainties, such as sea-state disturbances, renewable generation variability, and load fluctuations, into decision-making models. The surveyed studies can be broadly grouped into three methodological streams: stochastic modeling, robust optimization, and distributionally robust optimization (DRO). Robust optimization is typically built upon mixed-integer formulations and extended to a worst-case framework. By leveraging algorithms such as column-and-constraint generation, it ensures stable operation across disturbance realizations, making it well suited to scenarios with wave uncertainties induced by wind and currents or with uncertain PV power output [39]. In addition, some formulations embed component degradation costs into the scheduling model, thereby balancing long-term efficiency with operational reliability. In multi-energy shipboard power systems, robust modeling is also used to characterize PV uncertainty and to demonstrate that, even under worst-case conditions, the increase in operating cost can be limited while maintaining robust decisions [40].

Stochastic approaches place greater emphasis on probabilistic information. On the one hand, stochastic MPC is applied in operational settings, where uncertainties are incorporated into rolling-horizon prediction and optimization, and operating cost is jointly traded off against device aging [41]. On the other hand, stochastic programming and stochastic process modeling have also been adopted in broader maritime applications, offering representative paradigms for describing variability at its source [42]. DRO lies between stochastic and robust optimization: under uncertainty driven jointly by wind, waves, and PV generation, which may lead to speed loss and delay risk, DRO-based formulations have been proposed for coordinated generation and voyage scheduling. Compared with conventional robust approaches, these methods introduce partial distributional information through an ambiguity set to obtain less conservative solutions, while emphasizing cost optimization subject to on-time arrival and greenhouse-gas constraints [43].

From a literature-count perspective, robust optimization is applied across ship power management, shipboard power systems, and related maritime decision problems. Stochastic modeling covers both stochastic-MPC-based energy management and stochastic programming/process-based formulations, including posterior uncertainty representations. DRO further opens a pathway to examine the trade-off between incorporating distributional information and controlling conservatism [44]. Overall, uncertainty-aware energy management in this body of work can be summarized into three categories of modeling and solution paradigms: stochastic, robust, and distributionally robust approaches [40,45], shown in

Table 6. Comparative Overview of Uncertainty-Aware Energy Management Strategies for Electric and Hybrid-Electric Ships.

Representative Method	Vessel/Application	Validation Level	Computational Burden	Main Strength	Main Limitation
Robust optimization for generation–voyage co-scheduling	PEMFC + BESS electric propulsion under wind/current effects	Simulation	High	Provides reliable and cost-controllable scheduling across varying sea states and mission conditions	High modeling and solving complexity; conservatism may increase under wide uncertainty sets
Robust optimization for shipboard multi-energy power management	PV-enabled shipboard grids and integrated multi-energy dispatch	Simulation	High	Improves worst-case operational performance and supports coordinated dispatch under renewable uncertainty	Requires comprehensive system-level modeling and strong solver support
Stochastic modeling/stochastic MPC	Zero-emission FC/ESS/Shore-power EMS with weather uncertainty	Simulation/rolling-horizon case studies	Medium–high	Enables adaptive and aging-aware decisions under uncertain operating inputs	Strongly depends on the quality of the adopted uncertainty model and prediction framework
Distributionally robust optimization	Hybrid-ship microgrid dispatch and emission reduction under joint uncertainty	Simulation	High	Less conservative than classical robust optimization while retaining uncertainty resilience	High formulation complexity and limited evidence of broad real-time shipboard implementation

.

The analytical value of uncertainty-aware EMS lies in its ability to move beyond deterministic scheduling and to account more explicitly for renewable variability, load disturbances, and operational uncertainty. This makes such methods especially relevant for AES applications with fluctuating environmental and mission conditions. At the same time, the review indicates that the practical usefulness of these methods depends strongly on how uncertainty is modeled and on whether the resulting formulations remain computationally tractable for operational use. Accordingly, the evidence suggests that uncertainty-aware methods are highly valuable from a robustness perspective, but their broader deployment still depends on improved balance between modeling realism and implementation complexity.

3.4. Intelligent and Data-Driven Energy Management Strategies

The core of intelligent and data-driven energy management lies in transforming operational ship data into decision inputs that are predictable, optimizable, and monitorable, thereby supporting energy-efficiency improvement and operating cost control [46]. On the energy-consumption side, many studies take fuel/energy consumption forecasting as the entry point, using deep time-series models to capture temporal dependencies in voyage data. These works highlight the lagged nature of operational measurements and the critical role of temporal modeling in fuel-consumption prediction. By comparing RNN-based models, attention-based architectures (e.g., Transformer and Informer), and hybrid structures, they report that Informer achieves superior error and goodness-of-fit performance on real operational datasets, providing quantitative support for energy-efficiency management and emission-reduction objectives [47].

In scenarios closer to day-to-day operations, studies incorporate both at-sea and at-berth conditions into fuel-consumption modeling. By benchmarking multiple ML/DL models and hyperparameter configurations across different ship types, they indicate that the best-performing model and configuration vary by vessel and operating mode [48]. In addition, these works emphasize the contribution of meteorological features to at-sea prediction, reporting that incorporating sufficiently rich weather variables can improve at-sea forecasting accuracy by 4.49–16.46%, thereby offering actionable evidence for energy-saving practices in shipping companies and ports [49].

Beyond improving predictive accuracy, data-driven approaches are increasingly moving toward a “prediction–optimization” closed loop. Some studies develop fuel-consumption prediction models that balance accuracy and interpretability, use interpretability results to identify the key operating parameters to be optimized, and combine route segmentation with a parameter-optimization framework to produce operationally actionable energy-saving guidance [50]. Reported case studies on two vessels show energy-efficiency improvements of 6.41% and 7.05%, illustrating a practical pathway for translating interpretable deep learning into operating-parameter optimization [51]. In addition to consumption modeling, intelligent EMS also relies on state awareness and operational monitoring. For example, wavelet-based multi-scale frequency features can be fused with time-domain features, and combined with LSTM and KAN to perform short-horizon multi-step roll/pitch prediction, with uncertainty quantified via bootstrap to enhance robustness across sea states [52].

Another line of work integrates statistical thresholding with LSTM, using a “normal-behavior prediction–deviation thresholding” scheme to detect abnormal acceleration/deceleration, abnormal heading, and excessive deviation, thereby supporting intelligent monitoring systems and the safe operation of autonomous or semi-autonomous navigation [53].

From an analytical standpoint,

Table 7. Comparative Overview of Intelligent and Data-Driven Energy Management Strategies for Electric and Hybrid-Electric Ships.

Representative Method	Vessel/Application	Validation Level	Computational Burden	Main Strength	Main Limitation
Data-driven EMS via energy-use prediction	Fuel-use forecasting and sailing/Berthing management	Data-driven case studies/Simulation-based validation	Medium	Captures temporal dependence and provides predictive inputs for operational decision support	Mainly prediction-oriented rather than direct power allocation; strongly data-dependent
Multi-scenario fuel prediction	Company-level or port-level fuel management and emissions reduction	Multi-ship data analysis/Case-based validation	Medium	Broad comparative coverage and useful managerial insights across multiple operating scenarios	Best-performing model may vary by vessel and scenario, making transfer and deployment more difficult
Predict-and-optimize integrated framework	Operating-parameter tuning for fuel saving and emissions reduction	Case-based predictive optimization studies	Medium–high	Forms a closed loop from interpretable prediction to operational optimization with quantified gains	Multi-step workflow increases implementation complexity and integration burden
State prediction with uncertainty quantification	Sea-state-dependent roll/Pitch forecasting to support operation adjustment	Data-driven prediction studies	Medium–high	Provides forward-looking system states and uncertainty information to support adaptive strategy adjustment	Focuses on motion prediction rather than direct energy allocation; requires stable feature extraction and sufficient data support
Anomaly detection and condition-aware monitoring	Port and sea-trial monitoring, anomaly warning for (semi-) autonomous navigation	Monitoring-oriented case studies/Experimental data validation	Medium	Supports early fault/Anomaly detection and condition-aware operational monitoring	False alarms and noise discrimination remain important challenges in practical deployment

reveals that intelligent and data-driven EMS has expanded the field from optimization-oriented control toward prediction-enabled and adaptive decision support. Its main contribution lies in exploiting operational data to support functions that are difficult to realize using purely model-based approaches, such as anomaly detection, predictive scheduling, and data-assisted control adjustment. However, the review also indicates that the maturity of this method class remains uneven. Many reported advantages are conditional on dataset quality, vessel-specific operating conditions, and limited validation settings, while issues of reproducibility, transferability, and interpretability remain insufficiently resolved. Therefore, current evidence supports the strategic importance of data-driven methods, but not yet their uniform readiness for broad engineering deployment.

3.5. Cross-Method Comparison and Research Gaps

To facilitate intuitive and systematic comparison across method classes,

Table 8. Comparison Matrix of Major Energy Management Strategies for AES.

Method Classification	Computational Complexity	Real-Time Feasibility	Uncertainty Handling	Engineering Readiness
Rule-based	Low	High	Low	High
ECMS/A-ECMS	Low–Medium	High	Low–Medium	Medium–High
MPC/NMPC	Medium–High	Medium	Medium	Medium
Robust/Stochastic/DRO	High	Low–Medium	High	Low–Medium
Data-driven/AI-assisted	Medium–High	Medium	Medium–High	Low–Medium

presents a qualitative comparison matrix of the major AES energy management strategies reviewed in this paper. The matrix compares representative approaches using a common set of criteria, including computational complexity, real-time feasibility, un-certainty-handling capability, and engineering readiness. More broadly, the comparative discussion in this section also considers model dependence, data dependence, interpretability, and validation maturity. The purpose of this framework is not to assign exact numerical scores, but to provide a consistent basis for comparing strategy classes whose advantages and limitations are often discussed separately in the literature.

Overall, the reviewed studies indicate that no single strategy class is universally superior across all AES applications. Rule-based methods remain attractive for engineering deployment because of their simplicity, transparency, low computational burden, and strong real-time feasibility, but their adaptability is limited under highly dynamic multi-objective operating conditions. Optimization-based methods offer stronger constraint handling and more systematic coordination, yet their practical effective-ness depends heavily on model fidelity, parameter identification quality, and computational resources. In particular, MPC-type methods have shown promising results in many simulation and hardware-in-the-loop studies, while long-duration real-vessel validation remains comparatively limited.

Uncertainty-aware methods improve robustness under variable operating conditions, but usually at the cost of increased formulation and solution complexity. Da-ta-driven approaches are promising for prediction, monitoring, and adaptive decision support, but they face reproducibility and transferability challenges due to limited standardized maritime datasets, heterogeneous operating conditions, and uneven validation maturity. More broadly, the reported advantages of different EMS classes are often established under different underlying assumptions, and these assumptions strongly influence practical applicability across vessel types and deployment scenarios.

Taken together, the literature suggests that future research should place greater emphasis on transparent benchmarking, deployment-oriented validation, and methods that better balance algorithmic sophistication with practical shipboard implementability. In this sense, the practical value of an EMS cannot be judged solely by its theoretical or simulation-based performance. A method that is highly effective in a well-modeled and computation-rich research setting may be much less suitable in retrofit scenarios, in resource-constrained ship systems, or in environments with limited data availability. Future research should therefore move toward more deployment-oriented comparison, in which method classes are evaluated not only by control quality, but also by scalability, robustness, interpretability, and engineering feasibility across different operational contexts.

4. Applications of Energy Management Research in All-Electric Ships

With the rapid advancement of electric propulsion technologies, large-capacity Energy Storage Systems (ESS), and information and communication technologies, all-electric ships have gradually evolved from traditional single-power systems into highly coupled multi-energy integrated systems. Consequently, energy management and system optimization in electric ships exhibit characteristics such as increased structural complexity, diverse operational constraints, and pronounced multi-objective trade-offs. Against this background, the academic community has progressively established a systematic research framework encompassing system architecture configuration, energy management strategies, and coordinated operation among ships, ports, and microgrids.

4.1. Energy Management and System Optimization Under Multi-Energy Coupling

4.1.1. System Architecture and Configuration Optimization of Multi-Energy Ships

Existing studies commonly regard AES as “mobile microgrids”, in which diesel engines, fuel cells, renewable energy sources, and multiple types of energy storage units are integrated to provide unified power supply for both propulsion loads and onboard service loads. Within this framework, system architecture configuration and capacity matching are identified as key factors affecting energy efficiency, emissions, and operational reliability. Relevant studies indicate that multi-energy coupled architectures can enhance system redundancy while meeting dynamic power demand, thereby providing greater scheduling flexibility for the subsequent design of energy management strategies [54,55].

With respect to configuration optimization, the introduction of renewable energy sources and HESS has been demonstrated to effectively reduce fuel consumption and emission levels, driven by increasingly stringent emission regulations and the deployment of shore power infrastructure. From a techno-economic perspective, existing literature has analyzed the coordinated operation characteristics of energy units such as photovoltaic systems, diesel engines, and fuel cells, highlighting that appropriate capacity allocation and power matching play a significant role in system cost control and equipment lifetime extension [56]. Furthermore, incorporating energy storage sizing together with factors such as voyage route planning and shore electricity pricing into a unified optimization framework enables a comprehensive trade-off among cost, emissions, and operational performance under multiple sources of uncertainty [4,5]. In multi-storage coordinated architectures, the adoption of hybrid configurations combining fuel cells, batteries, and supercapacitors has been shown to effectively mitigate the impact of high-frequency power fluctuations on battery degradation and to enhance system operational stability under complex operating conditions [57]. In terms of powertrain configuration design, recent advances in electrified propulsion systems for three-dimensional transportation emphasize the importance of systematic parameter matching and topology selection [58]. Such approaches provide valuable guidance for the configuration optimization of multi-energy ship powertrains, particularly in balancing component sizing with operational requirements under complex marine conditions.

4.1.2. Evolution of Energy Management Strategies and Control Methods

As one of the key technologies for achieving decarbonization targets in the maritime sector, AES have witnessed rapid development in the field of energy management [59]. At the level of shipboard energy management strategies, existing studies have gradually formed a clear evolutionary trajectory from rule-based control toward optimization- and prediction-based control approaches. Early engineering applications predominantly adopted rule-based or threshold-based control strategies [60], which are straightforward to implement and impose low computational burdens. However, under multi-energy system configurations and complex operational constraints, such strategies struggle to simultaneously address multiple objectives, including economic performance, emission reduction, and energy storage degradation. To enhance decision-making performance, researchers have progressively introduced multi-objective optimization methods to achieve coordinated system operation.

Among optimization-based approaches, MPC has received considerable attention in recent AES research due to its receding-horizon structure and its ability to handle multiple constraints in real time. However, its practical advantages should be interpreted with caution. Control quality depends heavily on model fidelity, prediction accuracy, and solver speed requirements that may be difficult to meet in dynamic shipboard environments. Moreover, most existing evidence comes from simulation or hardware in the loop studies; long-duration validation in full-scale vessel operation remains limited. Therefore, MPC should be viewed as a promising but still case-dependent solution rather than as a universally applicable control paradigm.

4.1.3. Ship-Port-Microgrid Coordinated Energy Management

With the development of shore power infrastructure and port microgrids, energy management for electric ships has gradually expanded from single-vessel optimization to coordinated operation at the ship-port-microgrid level. Relevant studies have incorporated the berthing energy replenishment process of ships into the overall scheduling framework of port energy systems. Through coordinated optimization of onboard energy storage, shore power capacity, and port distributed energy generation, such approaches reduce the impact of berthing operations on the port power grid while lowering overall system operating costs [61].

Further studies, from the perspective of integrated energy systems, have unified the modeling of electrical and thermal systems and introduced battery temperature and safety constraints into the scheduling process, thereby improving the operational safety and economic performance of shipboard microgrids [62]. Meanwhile, in response to renewable energy intermittency and load uncertainty, intelligent control methods such as reinforcement learning have been employed to develop real-time energy management strategies. The results indicate that energy storage sizing and renewable energy deployment have a significant influence on berthing energy demand, providing quantitative support for the planning of port charging and battery swapping infrastructure [63].

4.1.4. Typical Application Scenarios and Progress in Engineering Validation

From an application perspective, multi-energy coupling and energy management strategies exhibit differentiated advantages across various operating stages. During port berthing, low-speed sailing, and mooring operations, assigning primary load supply to ESS and renewable energy sources can reduce the start–stop frequency of auxiliary diesel generators and lower noise and emission levels [64]. At the propulsion and auxiliary system levels, increasing the installed capacity of renewable energy can effectively suppress peak power demand and fuel consumption of the main engine [65]. For typical application scenarios such as harbor tugboats and service vessels, engineering validation results demonstrate that assigning primary load supply to battery–photovoltaic systems during berthing and low-speed operation enables the achievement of “zero-emission berthing” objectives. For medium- and high-power vessels, multi-energy coordination exhibits superior energy-saving performance under low- and medium-speed navigation conditions compared with single-energy configurations [66]. Furthermore, studies on fuel cell-battery hybrid propulsion systems indicate that, through appropriate design of fuel cell power ratings and energy storage capacity, near-zero-emission operation can be achieved in nearshore transportation scenarios [67]. From a microgrid perspective, by treating shipboard power systems as mobile offshore microgrids and combining scenario analysis with optimization methods, systematic evaluations of energy storage capacity allocation have been conducted for typical applications such as cruise ships and polar research vessels, providing quantitative guidance for energy system design under high-reliability requirements and extreme operating environments [68].

Overall, research on energy management for multi-energy ships has expanded from single-vessel operational optimization to integrated coordination among ships, ports, and microgrids. Nevertheless, there remains considerable room for improvement in unified modeling of infrastructure planning and operational scheduling, as well as in the characterization of service-level performance indicators.

4.2. Battery Safety Management and Lifetime-Oriented Energy Management

In multi-energy electric ship systems, the battery ESS constitutes a critical infrastructure for enabling electric propulsion and coordinated multi-energy operation. Research focusing on battery aging modeling, safety management, and scheduling under lifetime constraints has gradually formed a relatively systematic body of work.

4.2.1. Energy Management Strategies Considering Battery Degradation

By replacing conventional fuel engines with electric propulsion systems, electric ships offer advantages such as zero emissions, low noise, and high energy efficiency [69]. Owing to their high energy density and power density, lithium-ion batteries have become the most widely adopted energy storage technology in electric ships [70]. However, under operating conditions characterized by frequent start–stop events and rapid power fluctuations in ship propulsion systems, batteries are often subjected to high charge–discharge rates and significant thermal stress, resulting in pronounced capacity fade and internal resistance growth [71]. Existing studies have characterized the aging processes of marine lithium-ion batteries from both electrochemical mechanism-based and empirical modeling perspectives, identifying depth of discharge, C-rate, and temperature as key factors affecting battery lifetime. On this basis, several studies have incorporated indicators such as equivalent cycle count, capacity degradation rate, or state of health into system models. Considerable research efforts have been devoted to high-accuracy estimation of SOC, State of Health (SOH), and State of Power (SOP), and to reducing the risk of thermal runaway through thermal management, balancing control, and fault diagnosis mechanisms [72,73,74,75], thereby describing battery degradation characteristics over long-term operation. A recent review on shipboard battery state estimation highlights the unique challenges posed by shipboard operating conditions, including highly dynamic load profiles, marine environmental stresses, and stringent safety requirements [70]. These factors complicate accurate SOC and SOH estimation and underscore the need for robust algorithms capable of maintaining estimation accuracy under real-world shipboard scenarios.

From an energy management perspective, introducing battery lifetime constraints into energy management strategies, by limiting charge, discharge rates and optimizing power-sharing schemes, can significantly extend battery service life at the cost of only a marginal reduction in short-term economic performance [76]. In multi-storage coordinated scenarios, functional allocation strategies that combine batteries with other storage units such as fuel cells and supercapacitors have been demonstrated to be effective lifetime-friendly solutions [77]. In such configurations, supercapacitors handle transient power fluctuations while batteries are responsible for energy balancing, thereby effectively reducing the equivalent full cycle count of batteries [78]. Fuel cells provide stable output under low- and medium-frequency power demands, alleviating the long-term load burden on batteries. For high-power electric systems, studies incorporating battery lifetime models and equivalent cycle counting methods into capacity sizing have shown that appropriately increasing supercapacitor capacity can reduce the equivalent full cycle count of batteries [79]. Related research further indicates that, under high renewable energy penetration, increasing the ratio between battery energy capacity and power rating contributes to improved energy utilization efficiency and reduced unit energy consumption of the system [80].

4.2.2. Battery Safety Management and Coordinated Control with BMS

In addition to lifetime concerns, battery safety represents a critical factor constraining the large-scale deployment of electric ships. Battery Management Systems (BMS) ensure safe battery operation through functions such as state monitoring, thermal management, and fault diagnosis. With advances in data acquisition and modeling techniques, data-driven approaches have increasingly been applied to the online assessment and early warning of battery health states. In HESS, the coordinated design of BMS and EMS, whereby safety constraints are directly embedded into power allocation decisions, is widely regarded as an effective means of enhancing overall system safety and robustness.

4.2.3. Impacts of Uncertainty on Battery Performance and Energy Management

Under real-world operating conditions, battery systems in electric ships are subject to the combined influence of multiple sources of uncertainty. On the one hand, external environmental factors such as wind speed, wave height, ocean currents, and ambient temperature directly affect ship resistance, energy demand, and battery operating temperature and performance [21]. On the other hand, the stochastic nature of operational behaviors and load characteristics introduces additional uncertainty in power demand. For instance, the uncertainty associated with charging and discharging behaviors of electric vehicle users may degrade energy management performance, and similar challenges are observed in electric ships [81]. Moreover, battery aging processes themselves exhibit pronounced nonlinearity and limited predictability, further complicating lifetime assessment and energy management [82].

To address these challenges, some studies have employed scenario analysis, robust optimization, or clustering methods to enhance the stability and reliability of energy management strategies under uncertainty. The results indicate that, compared with conventional scheduling approaches, lifetime-oriented strategies exhibit superior performance in terms of safety margins and long-term operational effectiveness.

Overall, existing research has evolved from a primary focus on battery performance parameters toward a systematic framework integrating degradation modeling, safety management, and energy management strategies. Nevertheless, from the perspectives of engineering application and system planning, further efforts are required to strengthen the deep coordination between BMS and EMS and to systematically incorporate uncertainty factors into full life-cycle optimization models, thereby supporting the safe, reliable, and cost-effective operation of AES under complex operating environments.

A noteworthy gap in the existing literature is that battery degradation-aware energy management has advanced more rapidly in model construction and optimization formulation than in long-term maritime validation. Although many studies incorporate degradation cost, thermal constraints, or state-of-health indicators into control and scheduling frameworks, a considerable proportion of the evidence still comes from simulations, laboratory tests, or short-term case studies. Long-duration validation under real maritime operating conditions remains relatively limited, especially in the presence of route variability, sea-state disturbance, charging heterogeneity, and coupled operational uncertainty. This limitation should be recognized when interpreting claims regarding battery lifetime extension and safety enhancement.

4.3. Real-Time Energy Management Strategies Oriented to Engineering Applications

4.3.1. Rule- and Experience-Based Real-Time Control Methods

Limited by computational resources and modeling accuracy, early electric ships predominantly adopted rule-based or experience-based energy management strategies. These approaches feature simple structures and relatively strong robustness, and they are capable of satisfying basic operational requirements under conditions with controllable load fluctuations. To enhance adaptability under more complex operating conditions, researchers have introduced experience-based control methods such as fuzzy control and sliding mode control to improve dynamic performance and system stability [83].

4.3.2. Real-Time Energy Management Based on Optimization and Model Predictive

With improvements in shipboard power system modeling accuracy and onboard computational capability, optimization-based energy management methods have gradually been applied to real-time control scenarios. Among these methods, dynamic programming (DP) is often employed as an offline solution tool to obtain theoretically optimal control strategies for multi-energy systems under given operating conditions. However, due to its high computational complexity, DP is difficult to apply directly in online real-time control. Building upon this foundation, MPC is often regarded as a promising compromise balance between engineering feasibility and near-optimal performance through receding-horizon optimization and explicit constraint handling, and has gradually become one of the most actively studied optimization-based approaches in real-time energy management research [84]. Relevant studies have applied MPC to real-time power allocation in HESS and multi-energy shipboard power systems, enabling comprehensive optimization of economic performance, emissions, and energy storage degradation while satisfying operational constraints of batteries, fuel cells, and power grids. Comparative analyses with DP results indicate that MPC can closely approximate theoretical optimal solutions under most operating conditions, thereby validating its effectiveness in engineering applications. In addition, combining intelligent optimization algorithms with MPC can further enhance overall system performance while maintaining real-time implement ability [85].

However, these advantages should not be overstated. The real-time performance of MPC depends critically on the availability of sufficiently accurate prediction models, reliable forecasts of load and operating conditions, and computationally efficient optimization solvers. In real shipboard environments, model mismatch, environmental disturbances, sensor uncertainty, and hardware limitations may significantly affect controller performance. In addition, much of the current evidence remains concentrated in simulation and hardware-in-the-loop validation, whereas long-duration re-al-vessel demonstrations are still relatively scarce. Therefore, from an engineering perspective, MPC should be regarded as a promising but operationally constrained approach whose practical effectiveness remains strongly context-dependent.

4.3.3. Real-Time Scheduling Under Engineering Constraints and Application Scenarios

In practical engineering applications, real-time energy management strategies must explicitly consider operational constraints of shipboard DC distribution networks, including bus voltage deviations, short-term overload capability, and fault ride-through requirements. From the perspective of shipboard power systems, relevant studies have developed hierarchical control architectures that integrate fast dynamic response with slower energy optimization, thereby improving system stability and reliability under disturbance conditions [86].

In port berthing and energy replenishment scenarios, incorporating shore electricity prices, port emission constraints, and energy storage states into real-time scheduling models has been demonstrated to offer significant advantages in terms of both economic and environmental performance. Research findings indicate that coordinated operation modes combining shore power and energy storage can effectively shave peak loads and fill valleys, reduce the impact on port power grids, and lower emissions during ship berthing operations [87]. Such studies provide important engineering references for the coordinated operation of AES and port energy systems.

Overall, research on real-time energy management oriented toward engineering applications is evolving from single control strategies toward multi-level, multi-objective coordinated optimization frameworks [88]. In the future, with continued improvements in onboard computational capability, the maturation of intelligent algorithms, and the progressive enhancement of port infrastructure, real-time energy management strategies are expected to achieve higher levels of overall system performance while ensuring safety and reliability.

5. Existing Challenges for AES

Against the backdrop of the global shipping industry’s accelerated transition toward green and low-carbon development, AES are widely regarded as a key technological pathway toward near-zero or even zero emissions. However, despite substantial progress in research and engineering demonstrations, several critical challenges continue to hinder the large-scale deployment and commercial viability of AES. These challenges are organized below across three interconnected dimensions: system-level complexity, energy storage safety and uncertainty, and intelligent energy management integration.

5.1. System-Level Complexity and Lifecycle Co-Optimization

As AES evolve from conventional mechanically driven systems toward highly coupled “ship-port-microgrid” energy systems, their structural and operational complexity increases significantly. The deep integration of multiple energy sources, such as batteries, fuel cells, renewable generation, and shore power, introduces strong nonlinearities and multi-timescale interactions, which pose substantial challenges for accurate modeling, optimization, and real-time control. Existing studies often treat planning decisions (e.g., component sizing, route scheduling) and operational strategies as separate problems, leading to suboptimal system performance. Moreover, the lack of unified modeling frameworks that capture lifecycle cost, component degradation, and operational dynamics under uncertainty remains a critical gap. Addressing this challenge requires moving beyond the traditional separation of planning and operation toward integrated, lifecycle-oriented frameworks.

5.2. Energy Storage Safety, Reliability, and Uncertainty Management

ESS are the cornerstone of AES but also represent a major source of operational risk [89]. Complex maritime operating conditions, including prolonged high-power operation, frequent start–stop cycles, and harsh marine environments, significantly increase the likelihood of thermal runaway and system failures. While degradation-aware and safety-constrained energy management strategies have been proposed in the literature, their validation is predominantly limited to simulations, laboratory tests, or short-term case studies. Long-duration validation under real maritime operating conditions, incorporating route variability, sea-state disturbances, and charging heterogeneity, remains scarce. In addition, uncertainties arising from environmental disturbances, operational variability, and battery aging processes challenge the robustness of deterministic energy management strategies. The integration of battery safety management, uncertainty modeling, and EMS into a unified, practically deployable framework therefore remains an unresolved challenge.

5.3. Intelligence-Driven Energy Management and Green Port Synergy

With advances in artificial intelligence, digital twins, and next-generation communication technologies, ship energy management is shifting from rule-based control toward data- and intelligence-driven paradigms. However, the current application of intelligent algorithms often remains limited to individual ships or subsystems, with insufficient coordination with port energy systems, renewable infrastructure, and service networks. Key challenges include: a lack of standardized maritime datasets for model training; difficulty ensuring interpretability and robustness in safety-critical environments; and limited scalability of bespoke AI solutions across different vessel types and scenarios. Furthermore, the coordination between onboard EMS and port EMS remains fragmented, hindering the realization of system-level energy efficiency gains. Overcoming these challenges will require not only advances in algorithm design but also the development of interoperable architectures and validation protocols that support scalable, transferable, and safety-certifiable intelligent EMS.

6. Conclusions

This paper has presented a comprehensive and critical review of energy management technologies for AES. The reviewed literature shows that AES energy systems have evolved from relatively simple power-supply architectures into increasingly integrated multi-energy systems involving strong coupling among propulsion, generation, storage, voyage operation, and external energy interaction. As a result, shipboard energy management has moved beyond single-objective power allocation toward system-level decision-making across multiple time scales, multiple objectives, and multiple sources of uncertainty.

Across the existing literature, rule-based methods remain important because of their simplicity, transparency, and practical compatibility with shipboard PMS, especially in applications with limited sensing or computing resources. Optimization-based methods provide a stronger analytical framework for balancing fuel economy, emissions, battery degradation, and operational constraints, while MPC has become one of the most actively studied approaches for real-time coordination under constrained operating conditions. Uncertainty-aware methods further improve robustness in volatile environments, and intelligent/data-driven methods expand the scope of ship energy management by enabling prediction, monitoring, and adaptive decision support.

Nevertheless, the review also shows that many advanced methods are supported more strongly by simulation and pilot-scale validation than by long-duration full-scale vessel deployment. Their broader engineering adoption remains constrained by model dependence, data availability, computational burden, interpretability, and validation maturity. Building on these findings and the gaps identified throughout the review, the following future research directions are proposed as priorities for advancing AES energy management toward engineering-ready, scalable solutions.

• System-Level Lifecycle Co-Optimization. Current research predominantly focuses on operational energy management, while the coordination between planning decisions (e.g., component sizing, route scheduling) and real-time control remains fragmented. Future work should develop holistic optimization frameworks that integrate lifecycle cost, carbon footprint, and component degradation into a unified decision-making architecture. Such frameworks would enable systematic trade-offs between economic performance, environmental impact, and long-term system reliability.
• Energy Storage Safety, Reliability, and Uncertainty Management. Lithium-ion batteries remain a critical operational risk factor under complex maritime conditions. Although degradation-aware and safety-constrained EMS formulations have been proposed, their validation is largely limited to simulations or short-term laboratory tests. Future research should focus on long-duration, real-vessel validation of aging-aware and safety-constrained EMS, incorporating the full range of operational uncertainties—sea-state variability, load fluctuations, and battery state-estimation errors—into robust, stochastic, or distributionally robust optimization frameworks.
• Intelligence-Driven Energy Management and Green Port Synergy. The integration of artificial intelligence, digital twins, and next-generation communication technologies offers new opportunities for predictive and adaptive energy management. However, most intelligent EMS applications remain limited to individual vessels or subsystems, with insufficient coordination with port energy infrastructure. Future work should advance ship-port-microgrid co-management platforms that enable information sharing, coordinated dispatch, and real-time optimization across heterogeneous energy assets. Methods such as multi-agent reinforcement learning and digital-twin-enabled predictive control are promising directions for achieving adaptive, scalable, and resilient energy management.

In summary, while a well-defined technical architecture and methodological foundation for electric ship energy management have been established, widespread practical adoption hinges on strengthened interdisciplinary collaboration and system integration capabilities. As green port initiatives, smart shipping ecosystems, and digital infrastructure continue to evolve, energy management will assume an increasingly vital role in enabling large-scale electrification of maritime transport and building a truly sustainable and carbon-neutral shipping industry. The systematic understanding of energy management technologies provided in this review directly supports the global transition toward sustainable maritime transport by offering actionable insights for reducing emissions, improving energy efficiency, and integrating renewable energy sources into shipboard power systems.