A Review of Control Strategies for Brake Energy Recovery Systems

Abstract

Energy crises and environmental pollution continue to constrain sustainable development of the automotive industry. Large-scale deployment of electric vehicles (EVs) provides an effective pathway to reduce energy consumption and emissions. Regenerative braking technology plays a central role in improving energy utilization and extending EV driving range has sustained research attention. This study examines the operating principles, control strategies, and energy performance characteristics of regenerative braking systems (RBS). The historical development of brake-by-wire systems is reviewed, including system classification, structural configuration, and operating mechanisms, with an emphasis on their application in electric vehicles. On this basis, the working principles of regenerative braking systems are analyzed. A systematic review of regenerative braking control strategies is conducted across four categories: conventional control, fuzzy control, neural network control, and intelligent optimization algorithms. The analysis focuses on optimization methods for improving energy recovery efficiency and on the key factors governing energy transfer performance. Technical challenges associated with integrating regenerative braking systems into electric vehicles are further examined to provide a reference for future research and engineering development.

1. Introduction

As global energy constraints intensify and environmental pollution increases, enterprises worldwide are accelerating research on energy efficiency and low-carbon technologies [1,2,3,4]. New energy vehicles, particularly electric vehicles (EVs), have developed rapidly. Conventional energy recovery approaches based on hydraulic braking systems exhibit slow response and limited control precision, restricting their suitability for the diverse operating demands of modern new energy vehicles. To extend driving range and improve energy utilization, kinetic energy generated during braking is converted into electrical energy and stored in the battery, reducing overall energy loss. Brake-by-wire (BBW) energy recovery technology [5] has attracted considerable attention in heavy goods vehicles [6,7,8,9,10], urban buses [11,12], and rail [13] transport.

BBW [14] systems, as a new generation of intelligent braking systems, replace conventional mechanical or hydraulic actuation with electrical signal transmission, enabling faster responses and higher control precision. Their integration with energy recovery systems improves controllability and coordination, and provides a framework for multi-objective collaborative control. To maximize energy recovery efficiency while ensuring driving safety and ride comfort, effective and intelligent control strategies are required to coordinate torque distribution between electric and mechanical braking subsystems.

Various researchers have explored coordinated control of BBW systems [15] and regenerative braking systems [16]. Reported control approaches include rule-based strategies, fuzzy control [17], optimization algorithms (such as PSO [18] and GA [19]), model predictive control (MPC), and artificial intelligence methods including reinforcement learning and deep neural networks [20]. Key challenges remain, including limited real-time performance and strong subsystem coupling.

This study presents a systematic review of BBW technology in new energy vehicles. Operational principles, control mechanisms, and engineering applications are examined in detail. Future development trends of line control braking technology in the new energy vehicle sector are also analyzed, providing a theoretical basis and technical reference for improving braking performance and supporting industrial deployment.

The remainder of this paper is organized as follows. Section 2 reviews the historical development, structural configuration, and operating principles of line control braking systems. Section 3 introduces the architecture and working mechanisms of braking energy recovery systems. Section 4 discusses control strategies for optimized braking to improve energy recovery efficiency. Section 5 analyzes current challenges and future development trends. Section 6 summarizes the main conclusions.

2. An Overview of the Development of BBW Systems

2.1. The Evolution of BBW Technology

As illustrated in Table 1, automotive braking technology has evolved from mechanical systems to hydraulic, electronic, and intelligent integrated solutions. Early drum brakes featured simple structures but exhibited poor heat dissipation. The introduction of hydraulic actuation and disk brakes considerably improved braking performance. The widespread application of vacuum boosters in the 1960s reduced pedal effort. The subsequent adoption of electronic control systems led to the mass production of ABS (1978) and ESP (1990s), improving vehicle stability under extreme operating conditions. In the 21st century, EPB replaced conventional mechanical parking brakes, and drive-by-wire technologies such as EHB and EMB enabled faster system responses. Current braking systems are increasingly integrated with electric powertrains and autonomous driving technologies. Through regenerative braking and coordinated control, braking systems continue to advance toward higher levels of integration and intelligence.

Table 1. History of automotive braking technology.

Year	Technology Stage	Representative System/Application
1902	Drum Brakes	Mechanical Drum Brakes
1921	Hydraulic Braking	Lockheed Hydraulic Braking System
1953	Disk Brakes	Jaguar C-Type Racing Car Application
1960	Front Disk, Rear Drum + Servo	Vacuum-Assisted Distribution Structure
1992	ABS	Anti-Lock Braking System
1995	ESP	Electronic Stability Program
2001	EPB	Electronic Parking Brake
2010	EHB/EMB Line Control Braking	Electro-Hydraulic/Electromechanical Braking
To Date	Energy Recovery + Smart Integration	Autonomous Emergency Braking + Energy Recovery + Intelligent Control

2.2. Structure and Operating Principles of Cable-Actuated Braking Systems

With the progression of vehicle electrification and intelligent systems, braking architectures based on traditional mechanical or hydraulic transmission chains struggle to meet the requirements of high-precision control, rapid response, and energy recovery. Under these conditions, BBW systems [21] were developed. This technology replaces mechanical and hydraulic linkages with electrical signal transmission, enabling acquisition, processing, and execution of the driver’s braking command through electronic control. This architecture improves flexibility and controllability of the braking process. The core configuration typically includes a driver input module, an electronic control unit, actuators, and sensing and communication modules. In addition to conventional braking functions, BBW systems support regenerative braking, intelligent control, and autonomous driving features [22], providing integrated hardware and software support.

Cable-actuated BBW systems vary according to actuator type [23]. They are generally classified into hydraulic and mechanical configurations. The electro-hydraulic braking [24] (EHB) system represents a hydraulic drive configuration. Its structural layout resembles that of a conventional hydraulic braking system, retaining the hydraulic circuit while replacing the vacuum booster with an electrically driven assist unit. Owing to strong compatibility with existing platforms and minimal structural modification requirements, EHB has been adopted by major vehicle manufacturers as a transitional form of BBW technology. The electromechanical braking system (EMB) eliminates hydraulic transmission and employs electric motors to actuate the braking mechanism directly through electrical signals. As a fully electromechanical system, EMB provides rapid responses [25] and avoids brake fluid leakage. The removal of hydraulic transmission reduces the inherent response delay associated with conventional systems. Advances in control algorithms enable closer integration of EMB with longitudinal vehicle dynamics control systems, improving braking performance and coordination. Continued development of intelligent vehicle technologies is expected to accelerate the transition from EHB to EMB. Owing to its technical advantages, EMB has become a primary research focus for automotive manufacturers in China and internationally, and is expected to form the foundation of next-generation braking systems.

2.3. Electro-Hydraulic Braking System (EHB)

The EHB system, a representative form of BBW technology, comprises a pedal input unit, an electronic control unit (ECU), a hydraulic control unit (HCU), and brake actuators. During operation, pedal displacement and force are detected by sensors and converted into electrical signals transmitted to the ECU. Based on the driver’s braking demand, together with vehicle speed, load condition, and road surface state, the ECU determines the required braking force and regulates hydraulic pressure at each wheel through the HCU. This process enables precise and adaptive braking control [26]. Compared with conventional hydraulic braking systems, EHB improves braking performance while retaining the primary hydraulic architecture, including pipelines and control valves. The system incorporates redundant design to enhance operational safety. A key structural distinction lies in the brake booster configuration, where the conventional vacuum booster is replaced by an electrically driven booster or a pump-accumulator unit.

Figure 1 presents the basic structural configuration of the EHB system, and Figure 2 illustrates its operating principle. Figure 1 defines the interconnections among key functional modules. Figure 2 describes the dynamic coordination from driver input to control computation, and brake actuation. These diagrams provide the basis for subsequent system analysis and control strategy evaluation.

Figure 1. Schematic diagram of the basic structure of the EHB system.

Figure 2. Schematic diagram of the EHB system.

2.3.1. Evolution of EHB Systems

With continued technological development, electro-hydraulic braking systems have progressed from experimental prototypes to large-scale production. Multiple manufacturers have followed distinct development paths. In 2001, Bosch introduced an electro-hydraulic braking system based on SBC, marking the transition of EHB from conceptual design to practical application. In 2010, Bosch integrated electro-hydraulic braking with hybrid powertrains on the ESP HEV platform. In 2011, Audi introduced EHCB technology, validating a new generation of electro-hydraulic braking systems in passenger vehicles. In 2013, Bosch released the iBooster Two-Box configuration, followed by the more integrated iBooster One-Box architecture in 2015, improving system compactness and response performance. Continental launched the MK C1 integrated electro-hydraulic braking system in 2016 and upgraded it to the MK C2 platform in 2019, supporting electrified and automated driving applications. In 2017, ZF TRW introduced the IBC One-Box solution, further promoting global deployment of EHB systems. During the 2020s, industrialization accelerated. In 2023, Tainuva released the T-IBC One-Box system, highlighting the participation of emerging manufacturers into this field. EHB technology has evolved from distributed configurations toward highly integrated and intelligent systems, assuming an increasingly central role in electrification and autonomous driving applications. Table 2 summarizes the major BBW manufacturers and their representative products. The continued development of intelligent vehicle technologies is expected to promote the transition from EHB to EMB. EMB has attracted significant attention from global manufacturers, though large-scale application remains constrained by reliability limitations.

Table 2. Major BBW manufacturers and their representative products [27,28,29,30,31,32,33,34].

Year	Manufacturer	System/Product
2001	Bosch	SBC
2010	Bosch	ESP HEV
2011	Audi	EHCB
2013	Bosch	iBooster Two-Box
2015	Bosch	iBooster One-Box
2016	Continental	MK C1 One-Box
2017	ZF TRW	IBC One-Box
2019	Continental	MK C2 One-Box
2023	TAINUVA	T-IBC One-Box

2.3.2. Comparative Analysis of Performance Differences in Representative EHB Systems

With continued development of BBW technologies, manufacturers have introduced multiple EHB systems with distinct architectural features, performance characteristics, and application targets. Differences are observed in braking response time, pressure control accuracy, braking distance reduction, energy recovery capability, and redundancy design. These factors influence overall braking performance and are directly related to the energy management requirements of new energy vehicles, as well as the safety redundancy demands of autonomous driving systems. Table 3 presents a comparison of representative EHB products.

Table 3. Comparison of product performance differences.

	Braking Response Time	Braking Force Accuracy	Braking Distance	Fault Diagnosis	Weight and Volume	Energy Recovery Efficiency	Pedal Simulation
Bosch SBC	~150 ms	Error margin of 8–9%	Reduced by approximately 4 percent	Self-check function	Large in structure, complex in system	Limited	Feedback is somewhat lacking
Audi EHCB	~120 ms	Minimal 7%	Reduced by 7%	Enhanced fault detection mechanism	Compact in structure	Medium	Feedback is clear
Bosch iBooster Two-Box	~110 ms	Error margin of 5–8%	Reduced by 5–8%	Comprehensive fault monitoring functionality	Structure is somewhat oversized	Medium	Road feel is balanced
Bosch iBooster One-Box	~100 ms	Within 3% error	Reduced by 8–10%	Comprehensive diagnostic and protective functions	Compact and modular	Enhanced	Feedback is natural
Continental MKC1 One-Box	~90 ms	Approximately 2–3% error	Reduced by approximately 10 percent	Multi-level self-diagnostic and fault-tolerant protection	Compact design	Prominent	Feedback is consistent
ZF TRW IBC One-Box	~95 ms	Error less than 3%	Reduced by approximately 9 percent	Fault detection and alarm function	Medium structure	Stable	Feedback is accurate
Continental MKC2 One-Box	~82 ms	Error margin 1–2%	Reduced by approximately 12 percent	Real-time diagnostics and predictive maintenance	Compact in structure and lightweight	Optimal level	Feedback is smooth
TAINUVA T-IBC One-Box	~78 ms	Error less than 1%	Reduced by 15%	Fault monitoring coverage of 95%	Module integration height	Excellent	Feedback is genuine

Electro-hydraulic braking systems are commonly classified into Two-Box and One-Box configurations based on structural integration. These architectures differ in hydraulic module arrangement, control unit configuration, redundancy strategy, and system integration level. Table 4 summarizes the principal distinctions.

Table 4. Comparison of Two-Box and One-Box configurations.

	One-Box	Two-Box
Integration status	Integrated: incorporates ABS/ESP	Split system: independent of ABS/ESP
Structure	1 ECU, 1 brake unit	2 ECUs, 2 brake units
Cost	Highly integrated, relatively low cost	Low integration, relatively high cost
Safety	Pedal decoupling may pose a safety hazard	Pedal coupling can reduce safety hazards.
Energy recovery	High energy recovery efficiency, regenerative braking deceleration of 0.3–0.5 g	Low energy recovery efficiency, with regenerative braking deceleration below 0.3 g

2.4. Electronic Hydraulic–Electronic Mechanical Hybrid Braking (HBBW) System Operating Principle

EMB systems face practical constraints related to technical maturity and regulatory compliance. Installation near the wheel hub and brake assembly imposes stringent requirements on electronic components, thermal management, magnetic attenuation control, and environmental protection. These constraints limit large-scale deployment. The hydraulic–electromechanical hybrid brake-by-wire (HBBW) system combines EHB and EMB to support progressive industrial application of EMB technology. Current development focuses on a “wet-front, dry-rear” configuration. The front axle employs an EHB system to meet regulatory requirements for emergency braking. The rear axle adopts an electromechanical braking module to improve vehicle stability, reduce response time, and shorten braking distance. Figure 3 illustrates the operating principle of a typical wet-front, dry-rear HBBW configuration.

Figure 3. Schematic diagram of the HBBW system’s operating principle.

2.5. Electromechanical Braking System (EMB)

The EMB [35] represents a fully electromechanical BBW architecture that eliminates conventional hydraulic circuits. Braking force is generated directly by electric motors through mechanical transmission mechanisms. The system typically consists of a pedal input module, an electronic control unit (ECU), a motor drive module, and brake actuators. During braking, pedal displacement and force are detected by sensors and converted into electrical signals transmitted to the ECU [36]. The ECU determines the required braking torque based on driver input, vehicle speed, load conditions, and tire–road adhesion coefficients. Control commands are issued to the motor drive module. The motor generates torque and transmits it through mechanical elements such as gears, lead screws, or ball screws [37], converting electrical energy into clamping force that actuates the brake calipers and produces deceleration.

Compared with electro-hydraulic braking systems, EMB eliminates hydraulic pipelines and booster components, avoiding issues related to fluid leakage, hydraulic delay, and maintenance complexity. This simplified architecture improves packaging flexibility and energy utilization. Rapid motor response supports closer integration with regenerative braking systems and advanced vehicle dynamics control strategies. These advantages introduce higher requirements for motor reliability, electrical component durability, and thermal management. Redundant architecture and fault-tolerant control strategies [38] are required to ensure functional safety and stable operation under varying driving conditions.

To clarify system composition and operating mechanisms, Figure 4 presents the structural configuration of the EMB system, and Figure 5 illustrates its operating principle. Figure 4 shows the interconnections among key modules, including the pedal input unit, ECU motor drive module, and brake actuator. Figure 5 describes the signal transmission and execution process from pedal input to braking force generation [39]. These diagrams provide a basis for analysis of the system structure and control mechanisms.

Figure 4. Schematic diagram of the basic structure of the EMB System.

Figure 5. Schematic diagram of the EMB system.

Evolution and Industrialization Trends of BBW Technology in Automotive Applications

EMB represents a significant advancement in vehicle braking technology, driven by both technical progress and industrial development. The initial concept originated in aerospace and high-end automotive applications, where electric motors directly actuated brake calipers, demonstrating rapid response and high control precision. High development costs, stringent safety requirements, and extended reliability validation limited early adoption in experimental and small-scale demonstration stages. Increasing demand for new energy vehicles and autonomous driving has renewed interest in EMB. Application potential has been strengthened by independent wheel control capability, improved compatibility with regenerative braking, and direct integration with chassis domain controllers. In 2021, Brembo introduced the Sensify system, marking early industrialization exploration. In 2024, participation of domestic manufacturers such as Great Wall Motor and Jinyi Electronics Technology accelerated practical deployment. By 2025, international and domestic enterprises, including BETHEL Automotive Safety Systems, LeeKr Technology, Orient Motion, Continental, and ZF, expanded development of EMB products. Technical approaches included integrated brake caliper systems and wheel–hub motor drive configurations. Table 5 summarizes the representative manufacturers and their major products. These developments indicate a transition from extended technical accumulation and validation toward large-scale industrial implementation. EMB is expected to achieve broader vehicle integration in new energy and intelligent driving platforms, supporting further advancement of braking systems.

Table 5. Major EMB manufacturers and their representative products.

Year	Manufacturer
2021	Brembo—Sensify
2024	Great Wall Motor
2024	Jinyi Electronics Technology
2025	Bethel Automotive Safety Systems
2025	Orient Motion
2025	Continental
2025	LeeKr Technology
2025	ZF

2.6. System Characteristics and Performance Comparison of Electrohydraulic Braking and Electromechanical Braking

With the rapid development of new energy vehicles and intelligent driving systems, EHB and EMB have emerged as the two principal BBW architectures. Each presents distinct structural configurations, control characteristics, and application advantages. EHB enhances conventional hydraulic braking through electrically assisted pressure regulation, enabling improved brake force distribution and partial regenerative braking capability. EMB removes hydraulic transmission and achieves fully electric actuation, resulting in faster dynamic response. Key differences in response time, braking force precision, braking distance, and regenerative braking efficiency are summarized in Table 6.

Table 6. Comparative analysis of differences between EHB and EMB.

	Electro-Hydraulic Braking (EHB)	Electromechanical Braking (EMB)
Braking response time	Hydraulically transmitted, with a response time of 80–150 ms.	Direct motor drive, with response times as fast as 50–80 ms.
Braking force accuracy	Hydraulic control, affected by pipeline elasticity and hysteresis, offers moderate precision.	Closed-loop motor control with high precision, enabling finer-grained control.
Braking distance	Shorter than conventional methods, but constrained by hydraulic delay.	Shorter with more precise control, particularly demonstrating significant advantages in emergency braking and autonomous driving scenarios.
Brake energy recovery efficiency	Requires hydraulic coupling, with moderate energy recovery efficiency dependent upon control algorithm optimization.	Direct motor control delivers higher recovery efficiency and flexible energy management.
Advantages	Compatible with conventional braking architectures, enabling drive-by-wire adaptation without requiring significant modifications to chassis design.	High responsiveness and precision control.
Disadvantages	Braking performance is inferior to EMB, yet the hydraulic structure is retained, thereby increasing maintenance costs.	Technical requirements for motor performance and safety redundancy design are more stringent.

EHB and EMB represent two development paths in modern linear braking systems. EHB follows an incremental approach, improving control performance while retaining hydraulic compatibility, which supports transitional application in current vehicles. EMB adopts a fully electronic architecture characterized by rapid response and high precision, meeting the requirements of future intelligent driving systems. Differences remain in technical maturity, system costs, and reliability. Both technologies drive the evolution of braking systems toward higher efficiency and intelligence. Continued progress in electronic control, materials engineering, and functional safety design is expected to enhance synergy between EHB and EMB, supporting deeper integration of intelligent electric braking systems.

3. Brake Energy Recovery Systems

Brake energy recovery technology [40] is a core component of energy management in new energy vehicles [41]. The fundamental mechanism converts kinetic energy generated during deceleration into electrical energy for reuse, extending the driving range of electric vehicles [42]. During braking or coasting, the traction motor operates in generator mode. Part of the vehicle’s kinetic energy is converted into electrical energy and transmitted through the inverter to the power battery for storage. Energy that cannot be recovered is dissipated as heat through friction braking. Figure 6 illustrates the operating principle of the brake energy recovery system. Regenerative braking can extend vehicle driving range by approximately 25–40% [43], mitigating range limitations in new energy vehicle applications. Continuous improvement in motor characteristics, power electronics, and coordinated control strategies has promoted deeper integration between regenerative and friction braking, improving both energy recovery efficiency and system response performance.

Figure 6. Operating principle of brake energy recovery systems.

In practical applications, energy recovery in cable-actuated braking systems must balance efficiency, safety, comfort, and regulatory compliance. Electric braking and friction braking require coordinated allocation under varying operating conditions to maintain braking performance when regenerative capability is limited or unavailable. Pedal feel and braking smoothness must be preserved through pedal simulation and coordinated control to avoid abrupt deceleration changes. Energy recovery performance is constrained by motor characteristics, battery charge–discharge limits, thermal management capacity, and the vehicle energy management system [44], necessitating a dynamic balance between energy recovery and battery protection [45]. The regenerative braking process must remain compatible with stability control systems such as ABS and ESP, with braking safety prioritized during emergency maneuvers or low-adhesion road conditions. System design and implementation must comply with functional safety standards, including ISO 26262, and relevant regulatory requirements. Redundant braking circuits and reliability-oriented design ensure stable braking performance under diverse operating conditions.

Energy recovery in line-control braking systems requires coordinated consideration of safety, comfort, and efficiency in both design and application. Hardware improvements alone cannot satisfy these requirements. Control strategy development remains a central research focus. Key objectives include improving energy utilization while maintaining stable braking performance through optimized brake force distribution, adaptive regulation of regenerative intensity, and integration with overall vehicle control systems. Despite substantial progress further work is required in efficiency improvement, system coordination, and validation under real-vehicle conditions. The following section examines current control strategies and future research directions.

4. Brake Energy Recovery Control Strategies

Brake energy recovery control has evolved from conventional model-based approaches to intelligent optimization-based frameworks. Advancement of regenerative braking systems imposes stricter requirements on coordinated control, where energy recovery, braking stability, and driving comfort must be addressed simultaneously [46]. Early research relied primarily on vehicle dynamics models and ideal brake force distribution theory. Progress in artificial intelligence has promoted a transition toward adaptive, predictive, and optimization-oriented strategies. This section reviews four categories: classical control, fuzzy control, neural network control, and intelligent optimization methods.

4.1. Traditional Control Methods

Conventional control strategies have played a foundational role in regenerative braking systems for electric vehicles and hybrid electric vehicles (HEVs). Their advantages include structural simplicity, reliable real-time performance, and straightforward implementation. Coordinated allocation of regenerative and friction braking torque is typically achieved through vehicle dynamics modeling and predefined brake force distribution rules. This approach improves energy recovery while maintaining brake stability. Common strategies include controls based on the ideal braking force distribution curve (I-curve), proportional–integral–derivative (PID) control, and methods such as fixed proportional control, multistage control, and sliding mode control. These approaches monitor vehicle operating parameters, including speed, braking intensity, and battery state of charge (SOC) [47] and distribute brake torque according to predefined control laws. Such strategies maintain a balance between braking safety and energy recovery performance, providing the basis for subsequent development of adaptive and optimization-based control methods.

Conventional control methods, characterized by structural simplicity and strong real-time performance, form the foundation of regenerative braking systems. Their applicability is often constrained by rigid theoretical assumptions, hardware specificity, and limited adaptability to complex and time-varying environments. Ma [48] maintained system stability through fixed and continuous regulation, though the method was developed for compact front-wheel-drive electric vehicles and lacked scalability to other configurations. CS Nanda Kumar [49] balanced economic performance and stability in rear-wheel-drive commercial vehicles under an assumed equivalent motor power condition, which may deviate under battery aging or thermal derating. Liu [50] relied on predefined I-curve distribution, ECE regulatory constraints, and fixed adhesion coefficients, resulting in reactive rather than predictive control under variable road friction. Fixed ratio [51] and variable ratio [52] approaches improved stability under specific gradients such as UDDS downhill conditions, yet both depended on assumptions of ideal road surfaces and constant transmission efficiency, which limit robustness in real driving environments.

Economic-oriented controllers based on simple structures such as PID remain attractive due to low implementation cost. These strategies exhibit strong hardware dependence. Su [53] and Ashok [54] demonstrated improvements in energy economy, though each approach was developed for a specific system configuration. One study [53] assumed a hybrid energy storage system (HESS), while another [54] relied on torque characteristics of brushless DC motors with limited peak outputs. Such constraints restrict applicability primarily to electric motorcycles or light two-wheelers and limit transferability to full-scale passenger vehicles.

Advanced traditional strategies, including optimal control, hierarchical control, and slip rate control, improve precision yet commonly assume known environmental conditions. Sulakshan Rajendran [55] applied slip rate regulation during emergency braking and achieved effective torque control, though the method assumed ideal torque and voltage responses that may be affected by actuator delays and nonlinear dynamics. Gianfranco Rizzo proposed optimal [56] and hierarchical [57] approaches that maintained stability under wet and dry road conditions based on the assumption that the road adhesion coefficient or road state is known. In practice, real-time estimation of road adhesion is subject to uncertainty and measurement error. Related work [58] lacked experimental validation and adaptive learning capabilities. Several strategies were developed for specific vehicle categories, including hybrid vehicles with through-the-road (TTR) architectures [59], commercial vehicles equipped with EBW [57], and four-wheel EHB platforms, which limits cross-platform applicability.

Rule-based approaches such as the single-pedal strategy [60] aimed to improve driver feedback and deceleration coordination. These methods rely on fixed mapping rules and idealized motor efficiency models. A recurring limitation reported in [51,58,60] and Ref. [52] was the absence of experimental validation and adaptive learning capabilities. Unlike data-driven and adaptive control strategies, conventional control frameworks fail to maintain optimal performance over the vehicle lifecycle because of the lack of mechanisms that adjust to driver behavior, component aging, or environmental variation.

4.2. Fuzzy Control Strategy

Compared with conventional control approaches, fuzzy control strategies provide greater robustness and adaptability in nonlinear and multi-objective braking coordination tasks. Fuzzy control is based on linguistic variables and approximate reasoning, and has been widely applied in regenerative braking energy management. Multiple input variables, including vehicle speed, braking demand, battery state, and other operating parameters, are mapped into semantic categories such as low, medium, or high, simulating expert decision logic. The typical control framework is shown in Figure 7. A predefined fuzzy rule-based strategy performs multi-condition inference. The inferred results are converted into executable braking force distribution commands through defuzzification. This process enables coordinated allocation between mechanical braking and regenerative braking, maintaining a dynamic balance between energy recovery and braking stability. Configurable input variables and rule structures facilitate adaptation to nonlinear systems with strong coupling characteristics.

Figure 7. Fuzzy control system.

Fuzzy control improves robustness in nonlinear environments relative to traditional rule-based or model-based strategies. Its effectiveness remains influenced by subjective rule construction and assumed operating conditions. The absence of intrinsic self-learning capability limits adaptability compared with neural network-based approaches [61].

Hybrid strategies that integrate fuzzy controls with intelligent optimization algorithms [62,63,64,65,66,67,68,69,70,71] have been proposed to reduce subjectivity in rule design approaches, such as fuzzy–optimization coordination. Refs. [67,69] demonstrated stable performance under CLTC and NEDC driving cycles. These methods increase computational demand due to iterative optimization procedures, placing higher burdens on real-time control units. Many implementations rely on predefined constraints, including fixed braking intensity [63], constant vehicle speed assumptions [64], or preset safety boundaries [62,65,66,67,68,69,70]. Such assumptions may not reflect continuously varying real-world driving conditions.

A recurring limitation in many fuzzy control strategies is the oversimplified treatment of vehicle dynamics. Studies such as [72,73], and Ref. [74] assumed ideal road surfaces or considered only longitudinal dynamics, neglecting yaw motion and front–rear load transfer effects [72,75]. These assumptions reduce reliability during cornering, split-μ braking, or operation on uneven terrain. Applicability is often restricted to specific vehicle configurations. Approaches in [72,73] were developed for front-wheel-drive vehicles, Ref. [65] targeted parallel hydraulic hybrid platforms, and Ref. [39] was validated only under deceleration scenarios. These constraints limit generalization across diverse electric vehicle architectures.

Pure fuzzy control methods, including single fuzzy [72] and dual fuzzy structures [73], perform effectively in multi-variable coordination under standard driving cycles such as UDDS. Their rule bases remain fixed after design. Lack of adaptive learning prevents adjustment to component aging, parameter drift, or evolving driver behavior. Increasing rule complexity can improve simulation fidelity, as reported in [75], though this leads to larger model structures and higher implementation costs without addressing adaptability.

Several strategies focus on specific functions under constrained testing conditions. In [76,77,78,79] braking intent recognition was improved, though validation was conducted under limited operating scenarios with fixed speeds and predefined braking intensities. Many reviewed fuzzy control approaches [39,71,72,73,74,75,76,77,78,79] rely primarily on simulation environments such as NEDC or WLTC cycles. Limited experimental validation restricts assessment under real-world disturbances, including sensor noise, actuator delay, and rapidly varying road adhesion.

4.3. Neural Network Control Method

Neural network control differs from fuzzy strategies by replacing expert-defined rules with data-driven modeling. Through training on representative datasets, neural networks achieve self-learning and adaptive adjustment, enabling autonomous optimization of brake force distribution under complex operating conditions.

Neural network-based control methods leverage nonlinear mapping and adaptive learning to coordinate regenerative and friction braking in electric and hybrid vehicles. This approach improves energy recovery efficiency while maintaining brake stability. The general framework is shown in Figure 8. A typical structure consists of input, hidden, and output layers. Input variables include braking intensity, vehicle speed, battery state of charge, and pedal position. After training, the network generates brake force distribution commands for different braking scenarios, balancing energy recovery with vehicle stability.

Figure 8. Neural network control.

Neural network strategies rely on data-driven adaptation to address nonlinear vehicle dynamics. Their effectiveness is constrained by preset structural assumptions, simplified physical models, and algorithmic complexity that limits real-time implementation.

Many approaches depend on predefined mappings or operating states. In [80], adaptive fuzzy single-pedal control combined with neural networks achieved accurate deceleration tracking under UDDS, US06, and LA92 cycles. The control logic was based on a fixed single-pedal braking intent mapping, which limits flexibility when driver behavior deviates from predefined patterns. In [81], a GRBF–APSO–MPC hybrid structure improved smoothness during emergency braking based on an assumed preset driving style profile. Such assumptions reduce robustness across diverse driver characteristics. Hybrid architectures, including combinations of decision trees, LSTM networks, and metaheuristic algorithms [82] or neural networks and sliding mode control integration [83], emphasize braking stability and safety. These methods operate within predefined modes or safety boundary constraints, limiting responses under highly dynamic and uncertain conditions.

Model simplification is often introduced to reduce computational burden, which leads to validity limitations. In [84], an NN–MRC framework was applied for stability control, though validation was conducted using a single-wheel model with a fixed reference structure, neglecting full-vehicle dynamic coupling effects. In [85], an adaptive neural network was evaluated under the FTP-75 cycle, yet optimization remained constrained by adherence to the traditional ideal braking force distribution curve. Comparative analysis shows that both [80] and [85] were developed for front-wheel-drive configurations, without addressing rear-wheel-drive or all-wheel-drive dynamics [86].

Efforts to achieve higher precision often increase algorithmic complexity. In [83], a hybrid framework combined fuzzy logic, TD3 reinforcement learning, PSO–BP neural networks, and sliding mode control. The method incorporated tire modeling and braking intent recognition to maintain stability under normal and emergency conditions. Integration of multiple algorithms increased computational demand, creating challenges for real-time deployment on automotive electronic control units that require low-latency responses. In [80], multi-objective optimization similarly resulted in high computational complexity.

A common limitation across neural network-based strategies, including [80,81,82,83,84,85,86,87], is limited experimental validation. Most studies rely on simulation environments, such as UDDS and FTP-75, or on theoretical analysis. Some approaches [82,85] were not evaluated under standard driving cycles. This gap between simulation-based performance and real-world implementation restricts assessment of robustness under sensor noise, actuator delay, and variable road conditions [87].

4.4. Intelligent Optimization Algorithm

Unlike neural network-based optimization, intelligent optimization approaches emphasize global search capability and adaptive parameter adjustment. These methods may operate as direct controllers or as parameter tuning mechanisms to achieve multi-objective coordinated control in regenerative braking systems. Intelligent optimization algorithms have been widely applied in energy management for electric vehicles and hybrid electric vehicles (HEVs) due to their capability for global optimization, adaptive search, and effective handling of complex multi-objective problems. Common algorithms include the artificial bee colony (ABC), genetic algorithm (GA), particle swarm optimization (PSO), and sparrow search algorithm (SSA). These approaches simulate collective cooperation and evolutionary processes, iteratively refining candidate solutions to obtain optimal or near-optimal energy recovery control parameters. Integration of these algorithms enables coordinated improvement of energy recovery efficiency and braking smoothness under complex operating conditions, enhancing overall vehicle performance and energy utilization.

Intelligent optimization algorithms [71,81] (e.g., GA, PSO, ABC) demonstrate strong capability in global optimization and nonlinear constraint handling. Comparative analysis indicates that practical applications often involve a trade-off between computational cost and model specificity, and many studies lack experimental validation.

Genetic algorithms and hybrid variants exhibit strong stability performance but often introduce high computational demand. Studies such as [88,89] applied GA under NEDC and WLTC cycles. In [88], stability was achieved through optimization of a specific operating state and efficiency model, with limitations in computational efficiency and applicability restricted to front-wheel-drive (FWD) configurations. In [89], reliance on fuzzy rules and expert-defined knowledge constrained adaptability under uncertain or unstructured driving conditions. Hybrid strategies such as GA–PSO [90] and GA–ant colony [91] addressed multi-objective problems including adaptive cruise control coordination. These approaches increased algorithmic complexity, posing challenges for real-time implementation. In [90], validation was conducted under a single operating condition, without demonstrating robustness across multiple driving cycles.

PSO is widely adopted for its rapid convergence characteristics. General applicability is often limited by rigid model assumptions. Studies such as [92,93] emphasized fast convergence and Pareto optimality. In [92], the approach relied on specific control inputs. In [93], optimization was constrained by predefined braking force distribution logic. In [94], PSO was combined with PID control, though the framework remained theoretical without experimental validation. Methods reported in [39,95], as well as the firefly algorithm [96], achieved optimized brake force allocation under assumed drivetrain efficiency models. These formulations depend on predefined powertrain characteristics and require recalibration when applied to different configurations. In [94], PSO–PID integration was also presented without experimental validation.

The adaptive switching model-free adaptive control (AS-MFAC) approach [97] achieved speed tracking performance but was developed for a specific system model, which limits transferability. Advanced optimization methods aim to coordinate competing objectives such as braking safety and energy recovery. Many rely on idealized system representations. The artificial bee colony approach in [71] demonstrated adaptability under dry, wet, and icy conditions, with performance strongly dependent on braking intent recognition accuracy. Multi-objective frameworks such as NSGA-II [98,99] and TD-3 + DRL [100] addressed optimization [101,102] under CLTC and WLTC cycles. In [100], optimization was formulated around a dual-motor energy conversion model. In [98], braking force distribution followed fixed allocation logic. These structural constraints restrict broader applicability across different vehicle architectures.

A common limitation across these intelligent strategies [71,90,91,103] is the lack of experimental validation. Most studies rely on simulation environments such as NEDC and WLTC or on theoretical derivation, leaving real-world robustness and computational feasibility on vehicle hardware insufficiently verified.

4.5. Summary

Development of regenerative braking control algorithms has progressed from classical rule-based methods toward integrated and data-driven frameworks. Early model-based and rule-based approaches established the control foundation. Fuzzy control improved flexibility and robustness in nonlinear and multi-objective coordination through linguistic reasoning and expert-defined rules. Neural network approaches introduced data-driven learning and adaptive optimization under complex operating conditions. Intelligent optimization algorithms further serve as global search mechanisms to address trade-offs between braking safety and energy recovery efficiency. These strategies have evolved toward deeper integration, improving system intelligence, coordination capability, and energy utilization. Table 7 compares the energy recovery performance of representative brake force distribution strategies.

Table 7. Comparison of energy recovery efficiency for brake force distribution strategies.

Control Method	Core Mechanism	Test Conditions	Core Performance	Core Assumptions	Trade-Offs/Limitations
Basic Control Strategy	Ideal brake force distribution	——	Brake stability	Fixed and continuous Modify braking intensity	Compact front-wheel-drive electric vehicles
		Urban district high-speed M-IDC	Economy Brake stability	Equivalent motor power Exceeding braking requirements	Rear-wheel-drive commercial vehicles
		NEDC—High adhesion NYCC—High adhesion	Brake stability	Light–medium–heavy braking	Curve I ECE regulations Adhesion coefficient SOC
	PID control strategy	——	Economy	HESS-based braking	Electric motorcycles
		——	Economy	BLDC motor peak torque (10 Nm)	Electric two-wheelers
	Slip rate control strategy	Total of 30 m/s Emergency braking	Brake stability Torque regulation	Ideal gradient torque Slip rate voltage	Front-wheel hydraulic rear-wheel motor
	Optimal control strategy	Optimal braking operation Most severe braking	Optimum brake adjustment	Coefficient of adhesion is known	Hybrid vehicles (TTR)
		Three initial velocities Five types of braking distance	High computational efficiency	Ideal longitudinal force Known braking target	Lack of experimentation
	Tiered control strategy	Various road surface friction coefficients	Economical Precise torque control	Refine the pneumatic braking system	Commercial vehicles (EBW)
		Wet road surface Dry road surface	Brake stability	Known road conditions	Four-wheel hub electro-hydraulic brakes
	Fixed-ratio strategy	UDDS driving cycle—(2° and 7°) downhill	Brake stability Overcoming the lag effect in RLS algorithms	Specific longitudinal dynamics Specific motor Single good road surface	Lack of experimental data Lack of complex road surfaces
	Single-pedal + rule-based control	NEDC WLTC	Precise pedal travel feedback	Ideal road surface Ideal motor efficiency	Lack of experimentation Fixed mapping strategy Absence of adaptive learning
	Variable proportion brake Force distribution strategy	UDDS NEDC	Braking stability	Ideal road surface Transmission efficiency fixed	Lack of experimental data Lack of ramp scenarios
Fuzzy control strategy	Single fuzzy controller strategy	UDDS	Multi-dimensional decision-making	Ideal road surface	Front-wheel-drive vehicles Lack adaptive learning
	Fuzzy logic model	Suburbs Urban areas	Enhance simulation accuracy	Ignore yaw and front-rear axle distribution Full recovery	Large model size Lack of adaptive learning High costs
	Dual fuzzy controller	UDDS	Smooth braking	Ideal road surface Ignore lateral and roll dynamics	Lack of experimentation Insufficient adaptive learning Precursor vehicles
		NEDC EUDC	Cruise control	Ideal dynamics weighting introduction	Lack of experimentation Applicable to specific conditions
		WLTC CLTC	Cruise control	Braking force distribution safety boundary	Lack of experimental data Focus on deceleration scenarios
	Fuzzy control + intelligent optimization algorithm	NEDC FTP-75	Brake stability	Braking force distribution safety boundary	Lack of experimental setup Monotonous road conditions Limitations of subjective fuzzy controllers
		NEDC Light–moderate–heavy braking	Brake stability	Quantification of braking intensity Motor generator characteristics Specific road surfaces	Lack of experimentation
		Light braking Moderate braking	Path tracing	Brake force distribution Constant vehicle speed Simplified dynamic model	Lack of experimentation
		——	Brake stability	Considering only the relationship between vehicle speed and braking force	Lack of experimentation
		——	Brake stability	Ideal brake force distribution curve Fuzzy control rules	Lack of experimental data Lack of complex road surfaces
		CLTC-P	Brake stability	Considering only longitudinal dynamics Brake force distribution logic	Lack of experimental data Limited operational condition coverage Relatively complex algorithm
		UDDS CYC_1015	Smooth braking	Braking force distribution safety boundary tire model	Lack of experimental data Parallel hydraulic hybrid vehicles
		NEDC NYCC	Brake stability	Vehicle dynamics model	Lack of experimental data Algorithm computational complexity
		NEDC FTP-75	Brake stability	Vehicle dynamics model	Lack of experimentation
	Fuzzy control + braking intent recognition	Four distinct levels of braking intensity Single speed	Accurate braking intent recognition	Brake intent inference Control strategy logic	Operating conditions are limited Lack of experimental data
	Multi-mode + fuzzy control + algorithm	NEDC CLTC-P	Brake stability	Optimization algorithm objectives	Front-wheel-drive vehicles Lack of experimental data Complex calculations Limited operating conditions
Neural Network Strategy	Adaptive fuzzy single-pedal + multi-objective optimization neural network	UDDS US06 LA92	Brake deceleration tracking Brake smoothing	Single-pedal intent mapping	Lack of experimental data Computational complexity Front-wheel-drive vehicles
	NN–MRC with torque gradient control	UDDS Multi-surface emergency braking	Brake stability	Model reference architecture	Lack of experimentation Simplification of single-round models
	Neural network + synovial control	Low, medium, and high adhesion surfaces	Brake stability	Tire model Motor efficiency	Lack of experimental data Computational complexity
	C4.5 decision tree + LSTM neural network + seagull algorithm	——	Brake stability	Operating condition with preset braking force distribution safety margin	Lack of experimentation
	Adaptive neuro–fuzzy	FTP-75	Smooth braking	Ideal brake force distribution	Front-wheel-drive vehicles Lack of experimental data Lack of comparative analysis
	Fuzzy–TD3 + PSO–BP + SMC	Normal braking Emergency braking	Brake stability	Driving intent recognition tire model	Computational complexity Lack of real-world vehicle validation
	GRBF + APSO + MPC	Emergency braking (low adhesion, high adhesion)	Deceleration tracking Braking comfort	Driving style	Insufficient experimental coverage Inadequate operational condition coverage
Intelligent Strategy	Artificial bee colony algorithm	Dry, slippery, and icy road surfaces	Brake stability	Driver intent recognition	Lack of experimentation
	Genetic algorithm	NEDC	Brake stability	Optimization objective is singular Efficiency model is simplified	Front-wheel-drive vehicles Lack of experimental data Complex calculations
		NEDC WLTC	Brake stability	Fuzzy rule control	Lack of experimentation
	Particle swarm optimization	——	Brake stability algorithm convergence	Control input simplification safety logic	Lack of experimentation
		——	Brake stability	PID control	Lack of experimentation
		NEDC	Pareto optimal solution	Braking force distribution constraint	Lack of experimentation Single operating condition
		EUDC NEDC FTP-75	Optimization of brake force distribution	Braking force distribution constraints Power train system	Lack of experimentation Specific vehicle parameters
	Firefly algorithm	——	Optimization of brake force distribution	Braking force distribution constraint Power train	Lack of experimentation Specific vehicle parameters
	Adaptive control algorithm AS-MFAC	NEDC	AS-MFAC controller Speed tracking	System model	Lack of experimentation
	NSGA-II	NEDC CLTC WLYP	Multi-objective optimization	Braking force distribution constraint	Lack of experimentation
	TD optimization strategy 3 + DRL	CLTC WLTC	Brake stability	Dual-motor energy conversion model	Lack of experimentation
	Genetic + particle swarm	Speed adaptive follow Emergency braking	ACC follow Smooth braking	Layered architecture Multi-objective optimization	Lack of experimentation Computational complexity Single operating condition
	Genetic + ant colony	——	Optimization algorithms Multi-objective balancing	Multi-objective optimization	Algorithm complexity Lack of experimentation

Energy recovery efficiency depends on multiple interacting factors, including braking intensity, vehicle speed, front–rear brake force allocation, regenerative braking coefficient, control strategy, battery state, road adhesion, and motor–inverter efficiency. To maximize regenerative torque while maintaining braking stability, the front–rear braking force distribution coefficient K is constrained by the battery state of charge (SOC). The regenerative braking torque can be expressed as:

$$T_{reg}=\left\{\begin{array}{ll}{T}_{req}\cdot \eta \left(SOC\right),& ifSOC\le SO{C}_{\mathrm{m}\mathrm{a}\mathrm{x}}\\ 0,& ifSOC>SO{C}_{\mathrm{m}\mathrm{a}\mathrm{x}}\end{array}\right.$$

where T_req denotes the total required braking torque, $\eta \left(SOC\right)$ represents the regeneration efficiency coefficient associated with SOC [91], and SOC_max is the upper threshold that permits regenerative braking.

Appropriate coordination of these factors improves energy recovery and extends vehicle driving ranges.

Advanced control algorithms contribute to improved utilization of regenerative potential. Overall efficiency remains constrained by multi-physics coupling among mechanical, electrical, and thermal subsystems. Hardware configuration and mechanical characteristics define upper performance limits, including optimal matching of continuously variable transmission ratios (CVT) [104], brake disk material properties [105] and vibration characteristics that influence energy transfer stability [106,107]. Dynamic response of power and energy storage systems is also critical. Accurate modeling of permanent magnet synchronous motors (PMSM) [108], battery dynamics, engine engagement ranges [109], and coordinated management of hybrid energy storage systems such as supercapacitors [110] and fuel cells [111] directly affect recovery performance. Reliable estimation of battery SOC [112] is required for precise energy allocation. External operating conditions and driver interaction define boundary constraints. Traffic-informed route prediction [113], real-time speed variation patterns [114] and driver behavior characteristics [115] influence achievable recovery efficiency. Future research should focus on developing a unified optimization framework that integrates mechanical design, power system dynamics, control algorithms, and environmental factors to improve overall regenerative braking performance.

5. Challenges and Future Trends in BBW Systems

Regenerative braking technology has been widely deployed in new energy vehicles. Integration of regenerative braking systems into electric vehicles continues to present technical challenges, as illustrated in Figure 9. This section examines these challenges from four aspects: manufacturing and system integration, control strategies, perception and personalization, and safety redundancy. Development directions are outlined to support performance improvement and large-scale application.

Figure 9. Key challenges facing brake energy recovery systems.

• Manufacturing and Integration: Vehicle communication networks underpin information exchange. Deep integration of BBW and RBS requires high-speed, low-latency communication for coordination among multiple control units. Existing in-vehicle networks exhibit bandwidth constraints, electromagnetic interference, signal delay, and packet loss, which degrade coordinated braking and energy recovery control. Future architectures may adopt 5G NR-V2X communication, edge computing for local processing, adaptive routing protocols, network slicing for dedicated bandwidth allocation, and priority-based data transmission. These measures reduce latency and improve real-time responsiveness. Energy storage systems govern recovery efficiency and overall energy utilization. Inverters must maintain high efficiency and stability under frequent charge–discharge cycles. Batteries experience similar operating stress, affecting service life and safety [116]. Applications of SiC or GaN semiconductor devices, multilevel converter topologies, and integrated control algorithms can improve high-frequency performance. Development of high energy density materials and solid-state batteries can reduce thermal runaway risks and slow degradation. Hydraulic accumulators regulate hydraulic fluid in hybrid braking configurations. Sealing reliability and pressure stability remain critical. Under high-frequency braking and prolonged operation, fluid leakage can reduce braking force or cause failure. The use of high-temperature and corrosion-resistant materials, such as fluororubber and polytetrafluoroethylene, combined with multilayer sealing structures and optimized dynamic and static seal designs, can improve durability. Low-volatility and oxidation-resistant hydraulic fluids can further reduce degradation and leakage. Chassis space constraints require compact hardware integration. Increased integration strengthens system coupling and complicates thermal management. Future solutions may adopt highly integrated electronic modules and advanced packaging technologies such as three-dimensional integration and system-in-package (SiP) architectures to reduce wiring complexity and improve space utilization while maintaining thermal stability.
• System Control Strategy: Coordinated control of front and rear braking forces directly affects vehicle stability and energy recovery efficiency. During emergency braking, the system must ensure safety within minimal braking distance while maintaining energy recovery. Dynamic allocation between these objectives remains a core challenge. Control algorithms based on real-time vehicle states, including model predictive control (MPC) and deep reinforcement learning (DRL), provide viable solutions for regulating braking force distribution. Line control braking systems consist of multiple coupled subsystems. Interactions among subsystems and signal delays degrade overall control performance. Achieving coordinated control across subsystems remains a key issue. Compensation strategies for system latency, including delay prediction and dynamic adjustment, can improve stability and control accuracy under practical operating conditions.
• Perception and Personalization: In BBW architecture, the mechanical linkage between the pedal and actuator is replaced by electronic signal transmission. The loss of direct mechanical feedback reduces the driver’s ability to perceive brake force output, leading to inconsistent pedal feel. Force-feedback haptic systems combined with adaptive damping control can be applied to regulate pedal response in real time. Existing vehicle control systems provide limited personalization. Brake force distribution and energy recovery intensity are typically predefined and do not reflect individual driving behavior or comfort preference. Data-driven approaches, including machine learning and reinforcement learning, enable modeling of driver behavior and adaptive adjustment of control parameters while maintaining brake safety and energy recovery performance. Real-time parameter updating based on driver interaction feedback can further improve comfort and response consistency.
• Safety Redundancy: Both EHB and EMB architectures require redundant braking capability, such as standby hydraulic circuits in EHB and backup motor systems in EMB. In BBW systems, redundancy design ensures operational safety and fault tolerance. Current solutions are primarily derived from EHB and EMB configurations. Redundancy improves safety but increases structural complexity, system costs, and coordination requirements between primary and backup subsystems. Layered safety architectures that combine hardware and software redundancy provide a viable direction. Real-time system monitoring and diagnostic algorithms enable rapid fault detection and seamless switching between primary and backup subsystems. As braking systems evolve toward fully drive-by-wire and intelligent configurations, compliance with international safety regulations remains essential, including ECE R13 [117], ECE R13H [118], ECE R131 [119], ECE R90 [120], and FMVSS 203 [121], in alignment with national standards such as GB 13094-2017 [122].

6. Conclusions

Increasing energy demand and environmental pollution have accelerated the development of energy-efficient and low-emission vehicle technologies. Rapid expansion of the electric vehicle sector has promoted electrified and intelligent braking architectures. Regenerative braking systems play a key role in technological development and industrial deployment. This study reviews the existing literature [123] and examines the development trajectory of BBW technology [124] in new energy vehicles. System classifications [125], structural configurations [126], operational principles, representative products [127], and performance differences [128] were analyzed. Working mechanisms of regenerative braking systems and their control strategies were evaluated, including classical control, fuzzy control, neural network control, and intelligent optimization methods. Comparative assessment highlights the strengths and limitations of these approaches.

• Analysis of current technical challenges identifies key development directions for BBW technology. High-speed communication architectures supported by 5G and edge computing can improve system coordination. Applications of SiC/GaN devices and solid-state batteries can enhance energy storage performance. Advanced materials and sealing technologies can improve hydraulic reliability. High-density integration methods, including three-dimensional packaging, can address space and thermal management constraints.
• Deployment of dynamic control algorithms based on MPC or DRL enables coordinated optimization of braking safety and energy recovery. Latency prediction and dynamic compensation strategies can mitigate subsystem coupling and signal delays, improving system stability and response performance.
• Development of force-feedback simulation and adaptive damping control can improve pedal response consistency. Machine learning-based driver modeling enables personalized and adaptive optimization of braking energy recovery strategies.
• Integration of hardware and software redundancy within layered protection architectures strengthens fault tolerance. Real-time system monitoring and rapid fault switching improve operational safety. Compliance with international and domestic safety standards remains essential.

Continued progress in these areas will improve intelligence, integration, and energy efficiency in braking systems for new energy vehicles, supporting reliable intelligent driving systems and further advancement of autonomous vehicle technologies.