Solid-State Transformers in Modern Distribution Grids: A Comprehensive Review of Principles, Topologies, Key Technologies, Applications, and Challenges

Abstract

With the increasing complexity of distribution networks, higher demands have been placed on systems for efficient power conversion. The solid-state transformer (SST), which integrates power electronic converters with a high-frequency transformer (HFT), has become a major research focus in end-user power supply applications. This paper first compares the technical advantages of SSTs over conventional transformers and systematically explains their operating principles. It then reviews the development trajectory of SSTs in terms of topological evolution, prototype-based engineering validation, and the application of emerging materials. Next, it classifies and summarizes the current mainstream topologies and identifies core devices and key control technologies, including SiC devices and advanced soft magnetic materials. Finally, it introduces representative SST applications in data centers, smart grids, and charging stations and summarizes and discusses future research directions and challenges. This paper clarifies the technological evolution and existing bottlenecks of SSTs, provides a useful reference for the high-quality and highly flexible operation of distribution networks, and offers clear guidance and directions for the subsequent engineering deployment of SSTs.

1. Introduction

Conventional distribution transformers have undergone more than a century of development. Their manufacturing processes and performance have become relatively mature. Their capacity reaches the megawatt level, and their efficiency exceeds 99%. As a result, they have been widely used in power distribution and utilization systems [1]. With the continuous integration of distributed energy resources into modern power systems, users have placed higher demands on power supply reliability, flexibility, and power quality. Conventional distribution transformers can only provide voltage conversion between different levels and electrical isolation. They are therefore unable to meet the needs of flexible regulation in smart grids and high-quality power supply. They also still suffer from limitations such as large volume and single functionality [2,3]. Against this background, a new type of transformer based on power electronic conversion technology, namely the solid-state transformer (SST), has attracted widespread attention.

The SST, also referred to as the power electronic transformer (PET), is an intelligent power-conversion device. It relies on semiconductor devices, a high-frequency transformer (HFT), and advanced control techniques to convert high-voltage (HV) AC into low-voltage (LV) AC/DC and to enable bidirectional power flow [4,5]. Figure 1 illustrates the internal structure of the SST, consisting of the power cabinet and control cabinet. Compared with the conventional power-frequency transformer, the SST goes beyond the single operating mode based on electromagnetic induction. It integrates multiple functions, including electrical isolation [6], voltage conversion [7], and reactive power compensation. It therefore shows several notable features, such as a novel topology, flexible control, and a high degree of intelligence, as summarized in

Table 1. Comparison between conventional transformers and SST.

Characteristic Dimension	Conventional Transformer	SST
Working principle	Electromagnetic induction (50/60 Hz)	High-frequency (HF) power electronic conversion (≥1 kHz)
Volume/Weight	Large and heavy	Small and lightweight
Core function	Voltage conversion Electrical isolation	Voltage conversion Electrical isolation Active/reactive power regulation Fault current limiting Power quality regulation
Controllability	uncontrollable	Real-time and flexible control
Response speed	Slow (millisecond-level)	Extremely fast (microsecond-level)
Efficiency	High over a wide load range	Lower self-efficiency with system-level loss reduction potential
Fault handling	Reliant on external circuit breakers Slow breaking speed	Integrated intelligent protection Active isolation Fault current limitation

. Conventional transformers maintain high efficiency over a wide load range. While SSTs have lower device-level efficiency, their additional functions, such as power factor correction and unbalanced-current compensation, can reduce overall distribution-system losses. In addition, the SST is small and lightweight, has a reduced environmental impact, and offers a fast dynamic response. These advantages make it suitable for medium-voltage (MV) and HV, high-power applications. It can effectively meet the development needs of modern power systems and has become a key enabling device for future smart grids, renewable energy integration systems, and advanced power supply systems for emerging loads [8,9].

At present, SST research has attracted extensive attention both in China and worldwide [10,11,12]. Research institutions and companies around the world have invested substantial resources in key technological breakthroughs and have promoted their large-scale application. As a new type of intelligent energy conversion device, the SST can adapt to a wide range of application scenarios as well as power generation and supply systems. Its applications have become increasingly expanded into data centers, ultra-fast charging stations for electric vehicles, AC/DC hybrid distribution networks, and photovoltaic-storage DC flexible parks or microgrids [13,14,15]. Driven by the rapid growth in server power demand accompanying upgrades in computing infrastructure, the SST is expected to become a promising solution for next-generation power distribution systems in artificial intelligence data centers (AIDCs). In direct renewable-power supply scenarios, it can also enable efficient direct power delivery between renewable generation sources and load terminals, thereby reducing losses in intermediate conversion stages [16]. In addition to the functions of conventional transformers, the SST can also achieve real-time regulation of grid operation. This capability has attracted extensive attention from researchers and has promoted the increasing maturity of SST technology. As a result, the SST has shown strong application potential in fields that are highly sensitive to power quality, such as healthcare, banking, industry, defense, and aerospace. Their applications have also been gradually implemented in emerging fields such as rail transit and offshore wind power, bringing significant economic and social benefits for efficient energy utilization and green transformation [17,18].

Although SST technology has made substantial progress and its application scenarios have continued to expand, several urgent issues still remain in its practical deployment, including the application of new materials, control optimization, and cost reduction [14]. Against this background, a systematic review of SST topologies, core devices, and key technologies, together with a summary of current achievements and existing bottlenecks, has important theoretical and engineering value for promoting continuous technological breakthroughs and accelerating large-scale applications. Based on recent research progress and industrial practice in China and abroad, this paper presents a novel classification of SST development and provides a systematic review of SST topologies, core devices, control technologies, and emerging application scenarios. In particular, it distinguishes mainstream engineering topologies from emerging experimental designs and summarizes recent SST applications, offering a perspective beyond existing reviews.

2. Operating Principles and Development History of SSTs

2.1. Operating Principles of SSTs

The SST consists of two key components: a power electronic converter and an HFT. Its basic operating principle is shown in Figure 2. The power electronic converter mainly suppresses harmonics, phase, and frequency. The HFT mainly provides isolation between the primary and secondary systems and performs voltage-level conversion. With appropriate control strategies, the SST can effectively improve power quality [7,19,20]. The maximum allowable temperature rise in the core and windings, together with the saturation flux density of the core material, are the main factors that determine transformer size. For a given voltage and allowable flux swing, increasing the operating frequency can reduce the required core size and thus the transformer volume [21].

Researchers in China and abroad have proposed various SST topologies. According to whether a DC link exists in the overall topology, these topologies can be divided into two main categories. One is direct AC/AC conversion, in which no DC link is involved during the conversion process [10]. The other is AC/DC/AC conversion, in which a DC link is included during the conversion process [22].

AC/AC-type SSTs have the advantages of small size, light weight, and simple structure. However, their controllability remains limited. The secondary-side waveform largely reproduces the primary-side waveform, which makes it difficult to effectively control the power flow through the power electronic transformer [23,24]. In contrast, AC/DC/AC-type SSTs require more devices and have a more complex structure. Nevertheless, they possess more complete control strategies. By using PWM modulation, they can effectively control the primary- and secondary-side voltage, current, and power of the transformer. They can also ensure excellent power quality, and the primary-side power factor is adjustable. Therefore, AC/DC/AC-type SSTs have strong practical value and are likely to become a major focus of future research for power system applications [2].

2.2. Evolution of SST Technology

The evolution of SST technology has gone through four stages: theoretical exploration and concept establishment, architecture formation and prototype validation, engineering demonstration and scenario-based deployment, and device upgrading and performance iteration, as shown in Figure 3.

2.2.1. Stage of Theoretical Exploration and Concept Establishment

In the 1970s, the first HF-linked AC/AC conversion circuit was proposed, which is shown in Figure 4, marking the theoretical starting point of SST technology [25]. In 1980, the term “solid-state transformer” was formally introduced, which standardized the name of this technology [22]. Beginning in 1995, institutions such as the U.S. Navy and the Electric Power Research Institute in the United States carried out related studies on power conversion. They developed experimental prototypes based on AC/AC and DC/AC conversion and verified the technical feasibility of the concept. In contrast, these early prototypes were limited by device performance and still suffered from practical shortcomings, such as insufficient harmonic suppression and the lack of electrical isolation [26,40,41,42,43,44]. In 1999, a new topology consisting of an input stage, an isolation stage, and an output stage was proposed, marking the initial realization of the core SST architecture [22]. After that, the development of SSTs slowed because of limitations in power semiconductor devices. Even so, its theoretical framework basically took shape and laid the foundation for subsequent research.

2.2.2. Stage of Architecture Formation and Prototype Validation

In 2008, the National Science Foundation Engineering Research Center for Future Renewable Electric Energy Delivery and Management Systems (FREEDM) in the United States formally adopted SST as the name of its core device and clearly defined the standard three-stage architecture [29]. With technological breakthroughs and wider application of wide-bandgap (WBG) semiconductor devices such as SiC and GaN, the FREEDM Center at North Carolina State University began to explore the use of SiC devices in SSTs [27,28,29]. Figure 5 compares the evolution from its first-generation topology to its second-generation topology. In the second-generation SST based on 15 kV SiC MOSFETs, the overall structure was simplified, and the switching frequency was increased to 20 kHz [2,30,31]. By this stage, the basic SST architecture had taken shape. However, because the fabrication cost, packaging, thermal management, and control coordination of 15 kV SiC MOSFETs had not yet matured, SST development remained largely at the prototype stage.

2.2.3. Stage of Engineering Demonstration and Application

In 2014, ETH Zurich designed a 1 MVA SST to meet the demand for high-capacity SST in smart grids [32]. The rectifier stage and isolation stage adopted a diode-clamped three-level structure. This design increased the voltage withstand capability on the input side, greatly reduced the number of modules, and raised the switching frequency to 20 kHz [22]. In 2015, a 10 kV, 500 kVA industrial prototype developed by Professor Chengxiong Mao’s team at Huazhong University of Science and Technology completed trial operation in the Wuhan Iron and Steel grid. This was the first successful application of SSTs in an actual power system worldwide [45]. After that, research in China advanced rapidly. In 2018, Huazhong University of Science and Technology and the Institute of Electrical Engineering of the China Electric Power Research Institute jointly developed an MMC-based SST, as shown in Figure 6. In addition to conventional voltage transformation, it also provides power quality regulation. It is suitable for three-phase balanced load scenarios and has become a mainstream topology today [13,33,44]. During the same period, Zhejiang University proposed a topology integrating magnetic components and energy storage, which reduced photovoltaic power fluctuation to within ±5% [46]. As a result, SSTs have gradually moved from laboratory prototypes to field trial operation, and their engineering application and practical topologies have become increasingly mature.

2.2.4. Stage of Device Upgrading and Performance Iteration

Since the early 2020s, SSTs have developed toward higher switching frequencies, typically above 10 kHz. At the same time, third-generation semiconductor devices such as SiC devices have advanced rapidly, and HF, low-loss core materials such as amorphous and nanocrystalline materials have been widely applied in MV and LV AC/DC systems, distributed renewable energy systems, and emerging DC load hubs. These developments have steadily moved SSTs from the laboratory prototype stage toward applications featuring medium and high voltage levels, large capacity, and high power density [2,28,32,34,35,36,37]. In 2025, NVIDIA released the white paper “Building the 800 VDC Ecosystem for Efficient, Scalable AI Factories”, which identified the SST as a promising power supply solution for future artificial intelligence data centers [38]. This milestone marked a further expansion of SST application scenarios [39,40,46,47,48].

3. SST Topologies

According to the number of power conversion stages, SST topologies can be classified into three categories: single-stage [49,50,51,52,53,54], two-stage [54,55], and three-stage structures [54,56,57,58]. Among them, the two-stage structure can be further divided into two types, namely, configurations with the isolation transformer placed before or after the conversion stage, as shown in Figure 7.

Table 2. Comparative Analysis of SST Topologies.

Comparative Analysis of SST Topologies: Topology Classification	Single-Stage [49,50,51,52,53,54]	Multi-Stage
		Two-Stage [54,55]	Three-Stage [54,56,57,58]
			Cascaded H-Bridge (CHB)	Neutral Point Clamped (NPC)	Modular MULTILEVEL Converter (MMC)
Power/ Voltage rating	2–10 kVA 20–480 V	5–100 kVA 220 V–10 kV	1 kVA–3.3 MVA 39 V–5 kV
Efficiency Potential	High	Medium	Lower theoretical potential; high pratical efficiency possible with optimized design
Functionality and Control Flexibility	Low	Medium	High
Applications	Industrial power supply Motor drives On-board power systems Medium power scenarios	Traction power supply Smart distribution grids Distributed energy resource integration Medium-voltage distribution scenarios	Data centers Electric vehicle (EV) chargers Smart distribution grids and microgrids Military and industrial special applications
Advantages	Simple structure Easy modularization	Compact structure Good control performance	Mature control Scalable Simple module cascading	Simple control High power density Small filter volume	Modular High redundancy
Disadvantages	No DC link Limited voltage regulation	Great difficulty in modularization No MV DC-link protection	Independent DC bus Large device count	Limited voltage scalability	Complex control, Large device count
Cost	Low	Medium	Medium	Low	High

compares the characteristics and applications of different SST topologies and groups the two-stage and three-stage topologies into the multi-stage SST category for comparative purposes. The following section analyzes the typical SST structures.

3.1. Single-Stage and Two-Stage Topologies

Figure 8a shows a typical single-stage AC/AC SST proposed in [26]. This topology has a simple structure, high conversion efficiency, and bidirectional power transfer capability. Its power transfer capability is three times that of a power-frequency transformer [59].

Figure 8b shows the two-stage single-phase SST topology reported in [8]. In this topology, the isolation stage adopts a dual active bridge (DAB) converter, which directly rectifies high-voltage AC and steps it down to LVDC. However, the average active power transferred by this structure is highly sensitive to leakage inductance, the current fluctuation is large, and the regulation capability on the LVDC side remains weak [2,7,59,60,61].

The single-stage SST topology has limited functionality. The grid side does not provide power factor correction, so this topology is only suitable for special LV, low-power applications and cannot be used in MV distribution networks. Although the two-stage SST simplifies the overall structure to some extent, it requires twice as many modules.

Because the ripple current is relatively large, a larger smoothing inductor is required, and the achievable power factor is limited. In terms of controllability, the two-stage SST is also inferior to the three-stage SST. In addition, the two-stage SST cannot simultaneously achieve independent optimal control of the grid side, load side, and isolation stage. It is therefore difficult to adapt to modular MV cascaded topologies and multiport applications. For these reasons, it has become less favored in recent engineering practice and frontier research.

3.2. Three-Stage Topologies

The three-stage SST structure generally includes several main components, such as converters, a DC bus, and an HFT. Its typical configuration is shown in Figure 9 [62,63,64].

Although this structure is more complex than the AC/AC structure mentioned above, it provides more functions and a wider regulation range. For this reason, it has become the most widely discussed SST topology currently. The three-stage SST topology mainly has three typical forms: the cascaded H-bridge (CHB), the modular multilevel converter (MMC), and the neutral point clamped (NPC) topology [2,15,65].

3.2.1. CHB-Based Topology

The CHB-based SST topology is shown in Figure 10. This topology has a high degree of modularity, is easy to expand, and has relatively simple control strategies. Therefore, it has attracted widespread attention from researchers [66]. However, this structure requires a large number of converters and HFTs, which makes its configuration more complex when it is applied in distribution network scenarios [67,68,69].

3.2.2. NPC-Based Topology

The SST topology based on the NPC structure is shown in Figure 11 [13,32]. Compared with the CHB and MMC topologies, the NPC topology has only one HV input port, which can greatly reduce the number of HFTs required. However, this structure is difficult to expand in a modular manner. It also requires a relatively large number of clamping diodes, and voltage-balancing control is more challenging [70].

3.2.3. MMC-Based Topology

Figure 12 shows a typical three-stage SST topology based on the MMC structure [71,72]. Compared with the CHB structure, the MMC structure does not reduce the number of required power semiconductor devices. In contrast, it simplifies the internal control and gate-drive wiring of the submodules. Compared with the distributed layout of the CHB structure, it offers significant advantages in system integration and space utilization, which makes it more suitable for modular deployment in HV applications [70,73,74]. In addition, components such as the DC bus, DC sources, and DC loads in the MMC-based topology can be connected directly through the SST. This feature reduces the number of power conversion stages and improves conversion efficiency [75].

Because the three-stage SST has a DC bus, DC power sources such as photovoltaic generation systems can be directly connected to the system through the SST. This feature enables AC/DC hybrid integration and further improves energy conversion and utilization efficiency [8]. However, this structure is not well-suited for HV, high-power applications. A series connection of power devices on the HV side introduces voltage-balancing and reliability issues, which increase both cost and design difficulty. A comprehensive comparison of the CHB, MMC, and NPC topologies shows that the CHB topology has a simple structure, convenient scalability, and mature control. It also offers stronger redundancy and easier maintenance than the MMC topology, as well as better scalability than the NPC topology, without requiring a complex clamping structure. In MV distribution network scenarios, the overall performance of the CHB topology is more consistent with practical engineering needs. At the same time, the configuration of each stage in the CHB topology can be combined flexibly, which allows a variety of new topologies with stronger adaptability to be derived for different MV distribution network applications. Therefore, considering the characteristics of MV distribution networks and practical engineering requirements, the CHB topology has become the mainstream solution at present. It is more suitable for engineering deployment in MV scenarios and represents the engineering evolution direction of SST topologies in the MV field.

3.3. Emerging Topologies

At present, most emerging topologies are extensions of the three-stage SST. Figure 13 shows a cascaded multiport SST. By integrating a multi-winding HFT or parallel multi-module technology, it enables direct access to MV and HV grids while remaining compatible with various devices, including AC loads, DC sources, and energy storage systems. In this way, it serves as an energy router for AC/DC hybrid distribution networks [76,77,78,79]. In addition, this architecture combines the advantages of cascaded structures for MV and HV applications with the flexibility of multiport grid integration. It therefore performs much better than conventional transformers and standard SSTs in terms of system integration, power quality, operational reliability, and control flexibility. As a result, it has become a key development direction for the practical engineering application of SST technology.

The MMC-SST topology based on a switched-capacitor structure is shown in Figure 14. This topology consists of half-bridge MMC modules and a switched-capacitor interconnection circuit. The interconnection circuit includes a high-voltage AC bus, a synchronous transformer, and full bridges controlled by identical gate signals [80,81]. Voltage balancing of the submodule capacitors and the LVDC bus is achieved through the isolation transformer. Compared with conventional SSTs, the voltage ports in the isolation stage can adopt open-loop control. This advantage becomes more prominent when the number of LV ports is large, because the topology offers inherent voltage self-balancing capability [82].

In addition, the LF ripple currents of the submodules are also transferred to the high-voltage AC bus through the isolation transformer. Owing to the three-phase symmetry of the ripple currents, most ripple-current coupling is canceled naturally, which eliminates the need for dedicated ripple-current suppression and circulating-current suppression measures.

As summarized in Table 2, conversion efficiency is a key performance indicator for practical SST applications. The number of power conversion stages directly affects theoretical efficiency: single-stage SSTs require only one conversion and thus have the highest theoretical potential, two-stage SSTs introduce an additional isolated DC/DC stage with moderate efficiency reduction, and three-stage SSTs involve the most stages. However, practical efficiency is influenced by device voltage stress, high-frequency transformer design, soft-switching capability, and control complexity. Mature soft-switching techniques, such as DAB converters, and optimized modulation strategies enable high overall efficiencies in many medium- and high-voltage prototypes. Therefore, selecting SST topologies involves a trade-off between theoretical potential and practical feasibility.

A comparison of CHB, MMC, and NPC topologies reveals trade-offs among efficiency, control complexity, modular redundancy, scalability, and implementation cost. CHB topologies feature a clear structure, mature control strategies, and straightforward scalability, making them particularly attractive for medium-voltage applications. MMC offers superior modularity and high-voltage adaptability but requires more sophisticated control and communication systems, whereas NPC provides simpler control at the expense of limited scalability. Consequently, CHB remains one of the most mature and widely adopted SST topologies for medium-voltage applications. Emerging cascaded multiport SSTs and switched-capacitor MMC-SSTs are enhanced variants of conventional CHB- and MMC-based three-stage SSTs. Although prototype demonstrations have verified their feasibility, they remain largely at the experimental validation stage, while conventional CHB- and MMC-based SSTs have undergone more extensive engineering verification [77,79,80,81,82].

4. Core Components and Key Control Technologies of SSTs

As discussed above, topologies developed on the basis of the three-stage SST have become the current mainstream. However, further performance improvement and practical engineering deployment are still constrained by the performance bottlenecks of core components and the challenges of key control technologies. As shown in Figure 15, this section focuses on the core hardware components and key control technologies that support SSTs, with the aim of providing technical support for their engineering application.

4.1. Core Components of SSTs

Power electronic converters and HFTs are the two core components of the SST. The key supporting devices in power electronic converters are power semiconductors, and their performance directly determines the switching characteristics and energy conversion efficiency of the converter. The magnetic components inside the HFT serve as the core elements for magnetic energy storage and conversion, and they play a decisive role in the power density, loss level, and operational stability of the transformer. On this basis, the following sections discuss in detail the characteristics, performance comparison, and applications of these two core components.

4.1.1. Power Semiconductor Devices

Power semiconductor devices are the core switching units of the SST. Parameters such as voltage rating, switching frequency, power loss, and temperature directly determine the operating frequency, energy conversion efficiency, and overall design of the SST [83,84]. For a long time, silicon-based power semiconductor devices have served as the technological foundation of power electronics because of their mature fabrication processes and low cost. As a result, they have been widely used in various power conversion systems. However, as SSTs have placed increasingly higher demands on high switching frequency and high-voltage operation, the physical limitations of silicon-based devices have become more apparent. These devices can no longer fully satisfy the requirements of SSTs for high efficiency and compact structure, and they have become an important bottleneck that constrains further SST development [66].

Table 3. Comparison of different power semiconductor devices.

Characteristic	Si	SiC	GaN
Voltage rating	600 V–6.5 kV	650 V–15 kV+	200–650 V
Switching frequency	50–200 kHz (SJ MOSFETs) 20–50 kHz (IGBT/GTO)	100–500 kHz	≥1 MHz
Conduction loss	High	Low	Extremely low
Electron mobility (cm2/V·s)	1350	370–900	1700
Switching loss	High	Low	Extremely low
Thermal conductivity (W/cm·K)	1.5	3.2–4.9	2.5
Cost ($/A)	0.15–0.2	0.45–0.80	0.3–0.5
Application scenarios	Consumer electronics Power supplies Electric vehicles Industrial high-power converters	Electric vehicles Renewable energy EV charging	Consumer electronics Data centers Wireless power transmission HF applications

compares the characteristics of power semiconductors based on different materials [85,86,87,88].

Compared with conventional silicon devices, WBG devices such as SiC and GaN offer major advantages, including higher breakdown electric field, better thermal conductivity, and faster switching speed. These advantages can significantly reduce conduction and switching losses, increase the operating frequency, and decrease the size of the HFT [88,89,90]. Their excellent electrical and thermal performance aligns closely with the core requirements of SSTs for high switching frequency and low loss. As these devices continue to develop toward higher voltage ratings, more mature fabrication processes, and lower cost, WBG devices such as SiC and GaN are expected to become the preferred device choice for the engineering application of SSTs.

4.1.2. HF Magnetic Components

The HFT is the core component that enables high power density and HF isolation in the SST. Among its key elements, the core material has a decisive influence on transformer performance, loss, and power density [91,92]. A comparison of the properties of different materials is presented in

Table 4. Performance and application comparison of main magnetic core materials.

Category	Initial Permeability /(H/m)	Saturation Flux Density /(T)	Frequency Range /(kHz)	Power Range /(MW)	Advantages	Applications
Silicon steel	1000–4000	1.8–2.03	0.05–0.4	1–100+	Low cost High mechanical strength High saturation flux density	Power-frequency transformer Medium–low-frequency high-power converters Motor cores
Ferrite	<2000	0.3–0.5	50–1000	<0.1	Low HF loss Flexible shape	HF, Low-power applications
Amorphous & nanocrystalline alloy	<100,000	1.5–2	1–100	0.01–1	Good thermal stability Stable performance in large-sized magnetic cores	Medium-frequency high-power applications

.

For HV, high-power transformers operating from 400 Hz to 20 kHz, the candidate core materials generally include silicon steel, ferrite, amorphous alloys, and nanocrystalline materials [2,93,94,95,96,97]. Silicon steel has high permeability and high saturation flux density. However, its loss becomes very large at HF. Ferrite has relatively low loss and low cost, but its saturation flux density is low, which results in a larger core volume. Iron-based amorphous alloys can provide a saturation flux density as high as 1.56 T, and their loss remains at a moderate level. They are therefore suitable for HFT cores operating from several hundred hertz to several kilohertz. However, when the operating frequency reaches several tens of kilohertz, the loss can only be reduced by lowering the flux density. Another option is cobalt-based amorphous material, which has much lower loss than iron-based amorphous material, but it also has a lower saturation flux density. Therefore, when power density and loss are considered together, nanocrystalline material becomes the preferred choice. Its saturation flux density is much higher than that of ferrite, and its loss is the lowest among these materials. However, nanocrystalline materials are usually more expensive, and their raw material is generally supplied as a coiled strip. As a result, additional processing and treatment are required in practical transformer design.

4.2. Key Control Technologies of SSTs

The control system of the SST needs to be designed hierarchically according to the functional roles of the three-stage topology. The input stage is responsible for grid connection and power quality regulation. The isolation stage is responsible for power transfer and DC-bus voltage stabilization. The output stage enables adaptation to the LV grid and provides active support. The coordinated interaction among these control layers allows the system to meet multiple requirements related to efficiency, reliability, and grid services. This section systematically reviews the core control technologies for the three-stage SST topology. It clarifies the mainstream control strategies, technical bottlenecks, and development directions at each stage, thereby providing theoretical support for the engineering application of SSTs.

4.2.1. Input-Stage AC/DC Control Technology

Using the mainstream CHB topology as an example,

Table 5. Comparison of different control methods.

Control Method	Advantages	Disadvantages	Applicable Scenarios
Conventional PI/voltage, current injection control [99,100]	Simple implementation Mature engineering application	Slow dynamic response Poor adaptability to variable-frequency scenarios	Fixed-frequency HVDC Conventional grid-connected scenarios
SMC [101]	Fast dynamic response Strong disturbance rejection	Chattering problem Complex parameter tuning	Variable-frequency scenarios
MPC [104,105,106,107,108]	Multi-objective coordination Excellent dynamic performance	High computational burden Strict computing power requirements	High-performance grid connection Flexible DC scenarios
GFM (VSG/droop control) [109,110,111]	Provides inertia Supports grid stability	High control complexity Requires coordination of internal energy	High-penetration renewable energy grid connection Microgrid Islanded operation

compares different control methods [98]. Conventional linear control strategies have mature control structures, a high degree of engineering readiness, and stable performance under fixed-frequency conditions. However, they require targeted optimization under variable-frequency conditions [99,100]. To improve control performance in variable-frequency operation, sliding-mode control (SMC) has been widely applied to this type of converter. Compared with linear PI control, SMC provides better dynamic response and stronger disturbance rejection without increasing the computational burden [101]. It has therefore become an effective solution for variable-frequency scenarios. Model predictive control (MPC) has become a research hotspot in recent years because of its multi-objective optimization capability and strong adaptability to variable-frequency operation. However, the sharp increase in switching-state combinations caused by the cascaded multi-module structure of CHB leads to a substantial rise in computational complexity within the control loop, which limits its direct engineering application [102,103]. To address this issue, researchers have proposed lightweight strategies such as modulated MPC, switching-state preselection, multistage optimization, and predefined switching sequences. These methods reduce computational demand while effectively suppressing steady-state error, thereby promoting the practical application of MPC [104,105,106,107,108]. To meet the requirements for system inertia and primary frequency regulation under high penetration of renewable energy, grid-forming control has gradually been adapted to the CHB platform. By adopting grid-forming (GFM) strategies such as virtual synchronous generator (VSG) and droop control, the CHB-based SSTs can provide active voltage support, inertial response, and primary frequency regulation, effectively enhancing the stability and disturbance-rejection capability of the connected power grid [109,110,111]. It has therefore become a key issue in the design of this type of control strategy.

At present, conventional linear PI control remains the core strategy for engineering deployment because of its clear advantages in technical maturity and reliability. Lightweight model predictive control is rapidly moving toward practical implementation and is expected to become a mainstream next-generation solution. Grid-forming control, by contrast, represents the key development direction for power systems with high renewable energy penetration and is likely to become an essential control technology in future power systems. Overall, SST control is evolving toward higher performance, lower computational burden, and stronger grid-forming capability in order to meet the diverse demands of power-electronics-dominated power systems.

4.2.2. Isolation-Stage DC/DC Control Technology

The isolated DC/DC converter is the core stage in SSTs for achieving HF galvanic isolation and voltage conversion. Its main objective is to maintain the stability of the LVDC bus and to enable power transfer from the MV/HV DC side to the LVDC side. Among the mainstream topologies, DAB has been widely adopted. This topology consists of symmetrically configured fully controlled bridge legs and an HF magnetic coupling unit. By regulating the phase-shift angle between the primary- and secondary-side voltages of the transformer, it achieves precise control of both the magnitude and direction of the current [112]. This structure not only provides galvanic isolation between the primary and secondary sides but also establishes a bidirectional energy transfer path through HF carrier modulation [113].

From a modeling perspective, DAB dynamics are influenced by phase shift, transformer leakage inductance, port voltage ratio, load variations, and switching frequency. Average-value, small-signal, discrete-time, and generalized state-space models have been developed for controller design and stability analysis [114,115], supporting high-performance control and efficiency optimization.

Single-phase-shift (SPS) modulation remains the mainstream approach due to its simplicity and fast dynamic response, though it may induce circulating currents or reactive power under voltage mismatch or light load [116]. Multi-degree-of-freedom strategies, including dual-phase-shift (DPS), extended-phase-shift (EPS), triple-phase-shift (TPS), and asymmetric duty-cycle modulation, have been proposed to extend ZVS operation and reduce RMS current and reactive power [116,117].

In modular SSTs, multiple DAB cells require voltage balancing and power sharing control to prevent overvoltage, overcurrent, or thermal stress [118]. Efficiency can be further enhanced using extended soft-switching, minimum current-stress control, reactive power minimization, and coordinated parameter design [119].

Control architectures typically employ an outer-loop LVDC bus voltage controller combined with inner-loop power or current controllers. Beyond conventional PI control, advanced methods such as model predictive control (MPC), sliding-mode control (SMC), disturbance-observer-based control (DOB), adaptive control, and intelligent optimization have been applied to improve dynamic performance, robustness, and fault tolerance [120,121]. Closed-loop control is realized by a power flow controller, which takes the LVDC bus voltage and its reference value as inputs [36,62]. The operating characteristics of the DAB indicate that, under both forward and reverse power transfer modes, the system can be equivalently regarded as a combined inversion–rectification process, with a high degree of control flexibility [122].

At the lower level, control and modulation strategies are selected according to topology and operating conditions, including phase-shift, linearization, disturbance observer, and SMC approaches [123]. SPS control remains the practical engineering choice due to simplicity and fast response, while soft-switching techniques such as ZVS further improve conversion efficiency, reduce thermal stress, and extend device lifetime [61,124,125,126].

In summary, DAB-based isolated DC/DC converters have evolved into an integrated framework combining modeling, advanced modulation, voltage balancing, efficiency optimization, and multi-level control, achieving an optimized trade-off among efficiency, dynamic response, voltage/current sharing, reliability, and control complexity.

4.2.3. Output-Stage DC/AC Control Technology

The control mode of the LV-side DC/AC converter is determined by the application scenario of the system. It can operate in two basic modes: GFM and grid-following (GFL) [127]. In GFM mode, the converter needs to maintain the stability of its internal voltage phasor during the subtransient and transient periods. This function can be achieved through control methods such as VSG, dispatchable virtual oscillator control, and complex droop control [128,129,130,131]. In GFL mode, the control principle is similar to that used on the MV and HV sides. A current closed loop is adopted to achieve flexible regulation of active power, reactive power, and harmonic power. In addition, the LV grid control and service module can select signals such as active power deviation, frequency deviation, and voltage deviation as feedback variables or reference inputs according to operating requirements.

During grid-connected operation, GFL PI-based dual-loop current control is the dominant strategy because the structure is mature and highly reliable. In islanded and microgrid scenarios, VSG control and droop control of GFM have become the mainstream solutions, as they provide both inertia support and voltage–frequency stability. Owing to its diverse control schemes, scalability, and modular characteristics, the SST exhibits a high degree of flexibility. While performing the conventional functions of isolation and voltage transformation, it also provides dynamic grid services that are much faster than those of on-load tap changers, thereby meeting the diverse operational requirements of modern distribution networks.

5. Application Scenarios and Demonstration Projects of SSTs

5.1. Application Scenarios of SSTs

Owing to its efficient and flexible regulation capability, lightweight structure, and AC/DC conversion capability, the SST is mainly applied in high-density and high-reliability power supply scenarios. These applications include data centers, electric vehicle (EV) charging stations, renewable energy grid integration in future power systems, hybrid AC/DC distribution networks, microgrids, rail transit systems, and smart cities.

5.1.1. Data Centers

As computing infrastructure, data centers impose core requirements on their power supply systems, including high reliability, high energy efficiency, high power density, and high scalability. Conventional data centers usually adopt an AC distribution architecture based on an uninterruptible power supply (UPS) and a power-frequency transformer. However, this scheme has a complex structure, relatively low efficiency, large space occupancy, limited reliability, and high operation and maintenance costs. In addition, the 220 V AC architecture is not well-suited to distributed energy resources such as photovoltaics, energy storage systems, and fuel cells. The integration of multiple energy flows therefore requires additional power conversion equipment, which further limits system scalability. As shown in Figure 16, by replacing both the UPS and the distribution transformer, the SST integrates voltage conversion, galvanic isolation, power quality conditioning, coordinated energy storage, and multiport access into a unified platform. Through medium-voltage power supply and facility-level DC distribution, it simplifies the power supply architecture while simultaneously achieving high reliability, high efficiency, and high power density. As a result, it aligns closely with the core requirements of modern data centers.

5.1.2. Charging Stations

Conventional charging stations usually adopt architectures based on box-type substations, charging cabinets, or high-power DC EV chargers combined with charging interfaces. However, these solutions suffer from large volume, low conversion efficiency, and complex system architecture, and they are difficult to adapt to the capacity expansion required by ultra-fast charging [132]. As illustrated in Figure 17, the SST significantly reduces both size and weight while increasing efficiency to above 98%. It also supports millisecond-level power regulation and bidirectional energy flow. These features enable it to respond rapidly to time-varying grid demand and help achieve peak shaving and valley filling, making it well-suited to the requirements of modern charging and battery-swapping systems.

5.1.3. Smart Distribution Grids and Microgrids

In conventional renewable energy grid-integration schemes, distributed generators, energy storage systems, and DC loads all need to be connected to the AC bus through inverter stages. As depicted in Figure 18, the SST adopts a DC-bus-based architecture, which simplifies the system structure and enables flexible and rapid integration of distributed resources through DC interconnection. This approach supports efficient local consumption and surplus power feed-in, reduces operating losses, and improves the renewable energy hosting capacity [133]. In this context, the SST can serve as an energy router. It provides an efficient and intelligent interface for distributed energy sources such as solar and wind power, while also establishing a stable and reliable DC microgrid.

5.1.4. Military and Specialized Industrial Applications

Military sectors such as electrified transportation platforms, naval vessels, and aircraft impose stringent requirements on the size, weight, and power quality of power supply equipment. By using compact and weight-optimized SSTs for power transmission, these systems improve both design flexibility and energy utilization efficiency. In superconducting distribution systems for electric aircraft propulsion, the SST provides a compact and lightweight power transmission solution and increases design flexibility. In shipboard DC distribution systems, SST-based DC distribution improves overall energy efficiency.

5.2. Demonstration Applications of SSTs

Since 2011, SSTs have gradually entered engineering applications and practical deployment [39,95,134,135,136,137,138,139,140,141,142,143,144].

Table 6. Demonstration projects and industrial applications of SSTs.

Country/Region	Year	Rated Capacity/Voltage	Frequency	Conversion Efficiency	Application Scenario	Implementer
USA [134,135]	2011	1 MVA, 13.8 kV	20 kHz	97%	Navy laboratory AC/DC hybrid grid	GE and Wolfspeed
Switzerland [39]	2012	1.2 MVA, 15 kV	1.75 kHz	96%	Geneva railway system	ABB and Swiss Federal Railways
China [95]	2015	500 kVA, 10 kV	12 kHz	93.72%	Wuhan 10 kV power grid	Team of Prof. Mao Chengxiong, North China Electric Power University and Wuhan Iron and Steel (Group) Corp.
China [136,137]	2017–2020	3 MVA, 10 kV	/	97.89%	Suzhou Tongli AC/DC hybrid grid (EV charging station, data center)	State Grid Jiangsu Electric Power Co., Ltd. Suzhou Power Supply Branch
Korea [138]	2018	150 kVA, 13.2 kV	10 kHz	/	AC/DC hybrid grid	Sungkyunkwan University (SKKU)
USA [139]	2022	50 kVA/7.2 kV	16 kHz	97.8%	EV charging station, data center power supply	National Renewable Energy Laboratory (NREL)
Switzerland [140,141]	2022	166 kVA/7 kV	/	99%	Medium–high-voltage DC grid hybrid smart grid	ETH Zurich
Singapore [142,143]	2023	1.5 MW, 22 kV	/	98.5%	EV charging station	Nanyang Technological University (NTU), Singapore, Tanah Merah Control, National University of Singapore
China [144]	2024	2 MW/10 kV	14 kHz	98.3%	LVDC grid interface	Edon Research Laboratory

summarizes research and industrial projects worldwide that have advanced the development of SSTs.

SSTs are progressively transitioning from pilot demonstrations to large-scale applications. Their value has become increasingly evident in enhancing grid flexibility, facilitating renewable energy integration, supporting power supply for high-density loads, and advancing transportation electrification, making them one of the key enabling technologies for the future Energy Internet and next-generation power systems.

6. Challenges Facing SSTs

With their flexible topological structures and coordinated multiport control capability, SSTs have become one of the core enabling devices for driving the transformation of distribution networks toward higher efficiency and intelligence. However, to progress from laboratory-scale prototype validation to engineering implementation and large-scale commercialization, SSTs still need to overcome critical bottlenecks in both technological development and practical application [8].

6.1. Technical Challenges

The topological architecture of SSTs and their device-level operating characteristics subject them to reliability and lifetime challenges that are far more severe than those of conventional transformers. The main technical difficulties are concentrated in two dimensions: hardware and control.

At the hardware level, MV and HV SSTs generally adopt multilevel or three-stage topologies, in which a large number of power switching devices, capacitors, inductors, and control modules are integrated through the series connection of multiple cells and the parallel connection of multiple modules. As a result, the number of potential failure points increases dramatically, and a fault in any single device may cause system shutdown or even cascading failures. Meanwhile, although WBG devices such as SiC and GaN have provided the basis for HF operation, the switching losses generated at high frequencies, together with local hot spots caused by dense device packaging, have made thermal management a critical challenge. Existing air-cooling and liquid-cooling solutions still struggle to achieve precise temperature regulation under extreme operating conditions, thereby increasing the risks of device aging and thermal runaway. In addition, the multiport and multilevel characteristics of SSTs have led to fault modes that are increasingly diverse and strongly coupled. Fault signatures are easily disturbed by HF harmonics, while technologies for multi-fault coupled identification and early warning of weak incipient faults in complex topologies have not yet matured. The lack of a standardized fault diagnosis framework has therefore constrained the application of SSTs in scenarios requiring high reliability and long service life.

At the control level, the intrinsic nature of SSTs as “power electronic transformers” has fundamentally transformed their control strategies. Their multiport and multilevel characteristics impose extremely stringent requirements on response speed, robustness, and stability. In terms of control objectives, SSTs must simultaneously satisfy multiple tasks, including grid-side power quality regulation, precise load-side power supply, and energy management on the storage side. These objectives are strongly coupled, which makes coordinated optimization highly challenging. From the perspective of topology, multilevel and multimodule parallel structures have caused the controlled plant to exhibit high-dimensional, nonlinear, and strongly coupled characteristics, requiring effective solutions to issues such as capacitor voltage balancing and current sharing among modules. HF operation has also challenged conventional linear control methods, whereas advanced nonlinear algorithms still face bottlenecks related to high computational complexity and the difficulty of simultaneously ensuring accuracy and real-time performance in embedded controllers. Furthermore, operating conditions such as weak-grid impedance fluctuations and abrupt load variations have further increased the difficulty of control, and existing adaptive algorithms still need improvement in balancing dynamic response and system stability.

6.2. Application Challenges

The practical implementation of SSTs faces several key bottlenecks that limit large-scale commercialization, including high system cost, thermal management, electromagnetic compatibility (EMC), reliability, and high-frequency magnetic integration.

6.2.1. System Cost

High cost remains the primary barrier to SST commercialization, with the main contributors being power semiconductor devices and high-frequency transformers (HFTs). SST operation relies on wide-bandgap semiconductors such as SiC and GaN to achieve high-frequency, high-efficiency performance. However, the fabrication and packaging processes for these devices are complex, and the cost of individual devices far exceeds that of conventional silicon-based components. Moreover, multilevel topologies commonly used in medium- and high-voltage SSTs require dozens or even hundreds of power devices connected in series and parallel, significantly driving up system cost. HFTs, operating at kilohertz-level switching frequencies, require low-loss soft magnetic cores such as ferrite or nanocrystalline alloys, and windings typically use Litz wire to reduce skin and proximity effects. These materials and structures are substantially more expensive than silicon steel laminations and conventional copper windings used in power-frequency transformers. The high degree of customization and difficulty in mass production further reduces the cost competitiveness of SSTs.

6.2.2. Thermal Management

Thermal management is a critical challenge that directly affects power density, reliability, and device lifetime. High-frequency switching in SSTs leads to power losses concentrated in small chip-level regions, producing extremely high heat flux densities. The compact design required for high power density compresses available cooling space, making traditional air cooling insufficient. SSTs combine components with differing thermal characteristics, including power devices, HFTs, and filter capacitors, and multi-module cascaded structures can result in uneven thermal distribution and localized hotspots, accelerating device aging or triggering thermal runaway. Advanced cooling techniques such as liquid cooling, heat pipes, and immersion cooling can enhance thermal performance but increase system complexity, volume, and cost, creating a trade-off between thermal performance, compactness, and overall cost.

6.2.3. EMC

EMC is another major technical challenge in SST engineering. High dv/dt and di/dt generated by high-frequency switching, coupled with leakage flux from HFTs and surrounding metallic structures, are the primary sources of electromagnetic interference. EMC mitigation in SSTs remains challenging because high-frequency switching generates steep voltage and current transitions, and leakage flux from HFTs couples with nearby metallic structures. Conventional filters are limited in effectiveness, and improving suppression typically requires larger, heavier components, which conflicts with SST miniaturization and lightweight goals. The multiport architecture introduces multiple interference paths, while EMC characteristics strongly depend on topology, control strategy, switching conditions, and operating scenarios. Coordinated operation of multiple converters can further exacerbate system-level EMC issues, making comprehensive and standardized EMC design difficult.

6.2.4. Reliability and High-Frequency Magnetic Integration

SST reliability is inherently challenged by its complex structure and high-frequency operation. The system includes numerous power devices, capacitors, inductors, and control chips, and failure of any component can lead to system shut-down. Serial reliability decreases exponentially with the number of components, making multi-module cascaded medium- and high-voltage SSTs inherently less reliable than simpler low-frequency transformers. High-frequency operation accelerates device aging, and long-term reliability data for wide-bandgap devices under high-frequency and high-temperature conditions remain limited. Insulation and magnetic material degradation under high-frequency alternating fields are also not fully understood. The system’s resilience under grid faults or lightning strikes requires further study. High-frequency magnetic integration, which consolidates multiple inductors and transformers into a single magnetic core, can significantly reduce size and weight but introduces challenges such as magnetic coupling interference, complex multi-winding designs, and higher local losses, thereby increasing thermal management complexity.

In summary, the engineering deployment of SSTs requires comprehensive trade-offs among cost, performance, volume, reliability, and thermal and electromagnetic considerations. Future research should focus on cost-effective wide-bandgap devices, advanced EMC mitigation, efficient thermal management technologies, and high-reliability magnetic integration to accelerate SST maturity and enable large-scale commercialization.

7. Conclusions and Future Outlook

SSTs are driving the evolution of power systems from traditional passive networks toward greater flexibility and intelligence. This review has summarized their operating principles, topological variations, key component considerations, control technologies, and representative applications. Currently, SSTs are transitioning from laboratory prototypes to practical engineering demonstrations.

(1)

Device level

The widespread adoption of HV SiC and GaN WBG devices is enabling improved efficiency and power density under high-frequency, high-voltage, and high-temperature conditions. Three-stage SSTs, typically using CHB or MMC topologies, are widely adopted in medium-voltage prototypes, with several engineering demonstrations reported, while full-scale commercial deployment remains limited.

(2)

System level

Modular multi-level and multiport architectures, including MMC and CHB topologies, are approaching industrial feasibility, providing flexible AC/DC interfacing and power-flow control.

(3)

Control level

Conventional linear control has gradually been giving way to intelligent control. Control architectures based on model predictive control, reinforcement learning, and digital-twin technologies are being explored to improve multi-objective optimization, adaptive operation, condition monitoring, and predictive maintenance under complex operating conditions involving multiple ports, multiple objectives, and multiple time scales, while also enhancing the system’s capabilities for fault self-awareness, self-diagnosis, and self-recovery. These artificial intelligence and digital-twin-based methods, however, are still mainly in research and pilot implementation stages. SSTs are currently expanding from low- and medium-voltage scenarios into key application areas such as MV and HV distribution networks, data center power supply, and ultra-fast charging stations, where they are being explored as alternatives to conventional power-frequency transformers in selected scenarios.

In the future, as the cost of WBG devices declines further, magnetic integration technologies mature, and intelligent control algorithms are implemented in practical engineering systems, SSTs are expected to become key hub devices in next-generation power-electronics-based power systems under conditions of high penetration of distributed energy resources and increasing grid digitalization.