Research on Regenerative Braking Control Strategy for Single-Pedal Pure Electric Commercial Vehicles

: In recent years, with the increasing severity of energy and environmental issues, countries have vigorously developed the new energy automotive industry. To reduce the difﬁculty of driver operation and increase endurance mileage, this article proposes a regenerative braking control strategy for a single-pedal pure electric commercial vehicle. Firstly, the single-pedal control system’s hierarchical approach was designed to contain the driver’s intention analysis and torque calculation layers. After identifying the driver’s intention, a logic threshold method was used to determine the braking pattern. Then, a fuzzy theory was used, with road gradient, braking strength, and speed as input parameters, and the ratio coefﬁcient of braking force as the output parameter. A hybrid regenerative braking strategy was formulated based on the ideal distribution curve. Finally, the proposed strategy was veriﬁed through simulation and a constant-speed car-following experiment. The constant-speed car-following experiment results show that the maximum optimization rate of energy consumption provided by the proposed single-pedal regenerative braking control strategy is 5.81%, and the average optimization rate is 4.33%. This strategy can effectively reduce energy consumption and improve the economic performance of single-pedal pure electric commercial vehicles.


Introduction
With the reduction in oil supply and the increasing severity of global environmental issues, there is growing attention on electric vehicles [1,2]. Particularly in the field of commercial vehicles, electric commercial vehicles have emerged as a cost-effective option in the automotive industry. Currently, to alleviate environmental problems, some cities have banned fuel-powered trucks from entering urban areas, providing favorable conditions for developing electric commercial vehicles [3,4].
However, it is difficult to significantly improve endurance mileage due to technological limitations in electric motors and batteries. The regenerative braking system helps to recover kinetic energy through the vehicle braking process, generate electrical energy, store it in the battery, improve the endurance of pure electric commercial vehicles, and avoid energy (heat) dissipation caused by friction with brake pads, reducing the pollution released by the vehicle into the environment [5]. Therefore, improving regenerative braking efficiency is a critical focus in the research of electric commercial vehicles [6][7][8][9], and most researchers have focused on improvements in control strategies [10][11][12][13]. Ref. [14] proposed a hybrid powertrain regenerative braking energy recovery strategy based on fuel cells and supercapacitors, which could maximize braking energy recovery and reduce the energy consumption of braking resistors during the braking process. Ref. [15] divided the braking force of hybrid vehicles into three layers based on the total demand for braking force. The first layer distributed the total braking force between the front and rear axles; the second layer used fuzzy theory to allocate the braking force based on the regenerative braking ratio coefficient; and the third layer proposed a gear-shifting control strategy 2 of 16 based on the optimal economy for motor braking force and speed. Ref. [16] addressed the decrease in battery life caused by charging current during the regenerative braking process and proposed a hierarchical control strategy based on the battery aging model prediction. The studies as mentioned above on regenerative braking strategies are all focused on conventional dual-pedal vehicles.
The contact between car tires and different terrains has an impact on the thermal and wear states of the tires, and thus affects the efficiency of regenerative braking energy recovery. Ref. [17] established a regenerative braking control strategy that takes into account the variation of the tire-to-ground adhesion coefficient, improving the efficiency of energy recovery. Ref. [18] proposed a real-time estimation of road friction based on this model and the tire magic formula model to evaluate the potential friction, which improved the performance of autonomous vehicles in harsh environments. Ref. [19] integrates a nonlinear model predictive control (NMPC)-based ABS controller with tire thermal knowledge, which improves the performance of ABS. Ref. [20] proposed an electric all-wheel-drive regenerative braking system, which can often regenerate more braking energy compared to front wheel drive. The above research tends to weaken the impact of tire thermal and wear conditions on regenerative braking efficiency, while this article mainly studies improving the efficiency of regenerative braking by reducing the number of pedals.
The automotive industry has undergone a transition from manual transmission to automatic transmission. Undoubtedly, reducing the difficulty of driver operation is also one of the trends in automotive development. This has led to the introduction of single-pedal designs. Tesla is a well-known example of a single-pedal vehicle in practice. In this design, pressing down on the single pedal accelerates the vehicle while lifting the pedal applies the brakes, allowing for the seamless integration of everyday driving and regenerative braking. Meanwhile, an emergency brake pedal is also retained to ensure driving safety. In emergencies, drivers can use the emergency brake pedal to achieve rapid braking [21]. The single-pedal design simplifies and connects the acceleration and braking processes through intelligent adjustment of the electronic control system, improves energy efficiency, and provides a better driving experience for the driver.
The development of dual-pedal regenerative braking technology has laid the foundation for single-pedal regenerative braking technology [22]. Ref. [23] observed pedal torque demand through a new pedal dynamics model, which describes the dynamic changes in pedal position. A nonlinear model predictive controller was designed to achieve optimal energy management. Ref. [24] developed a single-pedal control strategy based on daily driving situations and established a single-pedal simulation model. Conventional braking conditions and urban driving cycles were selected for simulation, and dynamometer testing was conducted for verification. Ref. [25] focused on electric wheel loaders and proposed a single-pedal control strategy for electric wheel loaders to reduce the driver's workload during cyclic operations. Ref. [26] presented a single-pedal regenerative braking control strategy based on fuzzy theory, which allows the motor to generate braking torque based on the accelerator pedal to meet the driver's braking intentions. Overall, there has been limited research and practical implementation of single-pedal control strategies. Furthermore, most studies have just focused on software-in-the-loop and hardware-in-the-loop simulations, and fewer studies have involved experiment validation in the real world.
This study focused on single-pedal pure electric commercial vehicles. A fuzzy regenerative braking strategy was proposed based on recognizing driver intentions. The strategy was verified through simulation and a constant-speed car-following experiment in the real world. This research aims to reduce driver operation difficulty while reducing energy consumption and increasing endurance mileage.

Overall Framework of Single-Pedal Control Strategy
In this study, a hierarchical strategy formulation approach was adopted to establish the overall framework of the single-pedal control strategy for pure electric commercial vehicles, as shown in Figure 1.

Overall Framework of Single-Pedal Control Strategy
In this study, a hierarchical strategy formulation approach was adopted to establish the overall framework of the single-pedal control strategy for pure electric commercial vehicles, as shown in Figure 1.  The inputs include sensor signals, pedal aperture, and emergency brake pedal voltage. The VCU (vehicle control unit) processes these inputs through decision-making, resulting in outputs such as motor torque, motor speed, and brake actuation. After the decision-making process, the derived speed, torque, and brake force requirements are sent to the MCU (motor control unit), BMS (battery management system), hydraulic brake actuator, and other execution units. Simultaneously, the execution units provide feedback to the VCU, including actual motor speed, torque, and other monitored parameters.
This article takes a 4.5 t pure electric light truck as the research object, and the basic parameters of the vehicle are shown in Table 1.

Driver Intention Recognition
The recognition of driver intention involved three driving intervals during vehicle operation: the acceleration interval, the cruise interval, and the deceleration interval. Using pedal aperture variables and single-pedal vehicles' velocity, corresponding driving intervals could be recognized and partitioned. According to the practical experience of the The inputs include sensor signals, pedal aperture, and emergency brake pedal voltage. The VCU (vehicle control unit) processes these inputs through decision-making, resulting in outputs such as motor torque, motor speed, and brake actuation. After the decision-making process, the derived speed, torque, and brake force requirements are sent to the MCU (motor control unit), BMS (battery management system), hydraulic brake actuator, and other execution units. Simultaneously, the execution units provide feedback to the VCU, including actual motor speed, torque, and other monitored parameters.
This article takes a 4.5 t pure electric light truck as the research object, and the basic parameters of the vehicle are shown in Table 1.

Driver Intention Recognition
The recognition of driver intention involved three driving intervals during vehicle operation: the acceleration interval, the cruise interval, and the deceleration interval. Using pedal aperture variables and single-pedal vehicles' velocity, corresponding driving intervals could be recognized and partitioned. According to the practical experience of the team and after repeated testing and verification, we obtained the driver's intention function expression based on the actual vehicle motion state division. The interval separation curve is shown in Figure 2. team and after repeated testing and verification, we obtained the driver's intention function expression based on the actual vehicle motion state division. The interval separation curve is shown in Figure 2.

The deceleration interval
The cruise interval The formula for the boundary between the deceleration interval and the cruise interval is shown in Equation (1): The formula for the boundary between the cruise interval and the acceleration interval is shown in Equation (2): In Figure 2 and Equations (1) and (2), is the value of velocity and is the maximum threshold of velocity, and are the characteristic coefficients, denotes the value of pedal aperture, and , , and represent the maximum value of the pedal aperture corresponding to the acceleration interval, the cruise interval, and the deceleration interval, respectively.

Braking Classification Based on the Single Pedal
Section 2.2 divides the motion state of pure electric commercial vehicles into three intervals. Then, according to the practical experience of the team and after repeated testing and verification, the thresholds of pedal aperture, the difference between the current pedal aperture and the previous pedal aperture, the displacement change rate of the pedal aperture, and the road gradient are obtained. Therefore, in this section, the braking strategy during the determined braking interval is analyzed in detail using the logic threshold method. The braking strategy includes four modes: no braking, pure electric motor braking, pure brake actuator braking, and hybrid braking using an electric motor and a brake actuator. The steps are shown in Figure 3. The formula for the boundary between the deceleration interval and the cruise interval is shown in Equation (1): The formula for the boundary between the cruise interval and the acceleration interval is shown in Equation (2): In Figure 2 and Equations (1) and (2), v is the value of velocity and v max is the maximum threshold of velocity, m 1 and m 2 are the characteristic coefficients, pel denotes the value of pedal aperture, and pel am , pel cm , and pel bm represent the maximum value of the pedal aperture corresponding to the acceleration interval, the cruise interval, and the deceleration interval, respectively.

Braking Classification Based on the Single Pedal
Section 2.2 divides the motion state of pure electric commercial vehicles into three intervals. Then, according to the practical experience of the team and after repeated testing and verification, the thresholds of pedal aperture, the difference between the current pedal aperture and the previous pedal aperture, the displacement change rate of the pedal aperture, and the road gradient are obtained. Therefore, in this section, the braking strategy during the determined braking interval is analyzed in detail using the logic threshold method. The braking strategy includes four modes: no braking, pure electric motor braking, pure brake actuator braking, and hybrid braking using an electric motor and a brake actuator. The steps are shown in Figure 3.  In Figure 3, is the value of the single-pedal aperture, is the threshold of the single-pedal aperture, is the emergency braking pedal aperture, SOC stands for the battery state of charge, represents the difference between the current pedal aperture and the previous pedal aperture, _ is the given threshold of , represents the displacement change rate of the pedal aperture, _ is the given threshold of , is the road gradient, and stands for the given threshold of . The braking strategy based on the single pedal is as follows: Step 1: When the value of the emergency braking pedal aperture ( ) is more than 0, this indicates that the emergency brake pedal is being pressed. In this circumstance, only pure brake actuator braking is activated. Otherwise, go to Step 2.
Step 2: If the value of is equal to 0, the vehicle is not in an emergency braking state. Moreover, if the value of the single-pedal aperture ( ) is more than the given threshold value ( ), the vehicle is in an acceleration interval. In this case, braking is not needed. Otherwise, proceed to Step 3.
Step 3: When the battery state of charge (SOC) is greater than 90, in order to protect the power battery and prevent overcharging, regenerative braking is not applied, and only the brake system is activated. Otherwise, proceed to Step 4.
Step 4: When is greater than 0, indicating a positive pedal displacement, the driver's intention is determined to be acceleration, and no braking is applied. Otherwise, proceed to Step 5.
Step 5: When is equal to 0, if the road gradient is greater than the threshold value ( ), indicating an uphill road with a steep gradient, no braking is applied to ensure accurate driver intention recognition. If the road gradient ( ) is less than the threshold value ( ), only motor braking is applied to indicate a road with a moderate gradient. Otherwise, proceed to Step 6.
Step 6: When the road gradient is greater than the threshold value ( ), indicating an uphill road with a steep gradient, no braking is applied to ensure accurate driver intention recognition. Otherwise, proceed to Step 7.
Step 7: When the absolute value of pedal displacement (| |) is greater than the threshold value ( _ ), or the absolute value of pedal velocity (| |) is greater than In Figure 3, pel is the value of the single-pedal aperture, pel am is the threshold of the single-pedal aperture, pel emer is the emergency braking pedal aperture, SOC stands for the battery state of charge, x pel represents the difference between the current pedal aperture and the previous pedal aperture, x pel_max is the given threshold of x pel , v pel represents the displacement change rate of the pedal aperture, v pel_max is the given threshold of v pel , sl is the road gradient, and sl max stands for the given threshold of sl.
The braking strategy based on the single pedal is as follows: Step 1: When the value of the emergency braking pedal aperture (pel emer ) is more than 0, this indicates that the emergency brake pedal is being pressed. In this circumstance, only pure brake actuator braking is activated. Otherwise, go to Step 2.
Step 2: If the value of pel emer is equal to 0, the vehicle is not in an emergency braking state. Moreover, if the value of the single-pedal aperture (pel) is more than the given threshold value (pel am ), the vehicle is in an acceleration interval. In this case, braking is not needed. Otherwise, proceed to Step 3.
Step 3: When the battery state of charge (SOC) is greater than 90, in order to protect the power battery and prevent overcharging, regenerative braking is not applied, and only the brake system is activated. Otherwise, proceed to Step 4.
Step 4: When x pel is greater than 0, indicating a positive pedal displacement, the driver's intention is determined to be acceleration, and no braking is applied. Otherwise, proceed to Step 5.
Step 5: When x pel is equal to 0, if the road gradient is greater than the threshold value (sl max ), indicating an uphill road with a steep gradient, no braking is applied to ensure accurate driver intention recognition. If the road gradient (sl) is less than the threshold value (sl max ), only motor braking is applied to indicate a road with a moderate gradient. Otherwise, proceed to Step 6.
Step 6: When the road gradient is greater than the threshold value (sl max ), indicating an uphill road with a steep gradient, no braking is applied to ensure accurate driver intention recognition. Otherwise, proceed to Step 7.
Step 7: When the absolute value of pedal displacement (|x pel |) is greater than the threshold value (x pel_max ), or the absolute value of pedal velocity (|v pel |) is greater than the threshold value (v pel_max ), indicating a high driver expectation for braking, motor braking alone cannot provide sufficient braking force. In this case, both motor braking and brake system braking are applied, resulting in a hybrid braking mode. Otherwise, only motor braking is applied.

Formulation of Single-Pedal Regenerative Braking Strategy
In a single-pedal electric vehicle, regenerative braking is achieved through the motor. Therefore, this study focuses on two braking modes: pure motor braking and hybrid braking, based on the driver's braking demand.

Regenerative Braking Principle
Regenerative braking is a braking method that utilizes the electric motor of an electric vehicle to convert kinetic energy into electrical energy and store it. Its principle is that during the braking process, the electric motor is used as a generator, converting the vehicle's kinetic energy into electrical energy, and storing the electrical energy in the battery. When the single-pedal signal is in the braking state, the electric vehicle's electronic control system will receive the braking signal and switch the electric motor to the generator mode. At this point, the electric motor starts to operate, and the movement of the wheels drives the motor rotor to rotate, generating electrical energy. This electrical energy will be rectified and regulated by the electronic control system, and then stored in the battery.

Formulated Based on the Single-Pedal Braking Strategy
Based on the single-pedal signal and the current motor speed, the driver's demand for car braking force can be obtained. When the driver's demand for car braking is low, only motor braking is required. The road gradient, pedal displacement, and pedal displacement rate are used as input variables for motor braking. The motor's regenerative braking torque is determined by applying certain weight ratios. When additional braking force from the brake system is not required, the regenerative braking torque is calculated as follows: In this equation, T m represents the current regenerative braking torque, α is the road gradient conversion coefficient, β is the pedal displacement rate conversion coefficient, γ is the pedal displacement conversion coefficient, and α, β, γ ∈ (0, 1). v pel is the current pedal displacement rate, x pel is the current pedal displacement, v max is the threshold pedal displacement rate for increasing the brake pedal displacement, x max is the threshold pedal displacement for increasing the brake pedal displacement, and T m−max is the peak torque of the motor.

Hybrid Braking
When the driver has a high demand for car braking, both motor and brake braking are required to participate in the braking process. To ensure the handling stability of the car during braking, it is necessary to distribute the braking force between the front and rear axles based on the ideal distribution curve.
The main factors affecting the efficiency of vehicle braking energy recovery include braking intensity and vehicle speed [27]. Pure electric commercial vehicles usually drive on mountainous terrain, and so road gradient is also an important consideration. These three factors can be obtained in real time using sensors and measuring devices and are used as input variables of the fuzzy controller. The recovery of braking energy mainly depends on the amount of motor braking torque involved. During the braking process, it is necessary to adjust the distribution ratio between the motor braking and the brake braking to achieve the best energy recovery efficiency. This distribution ratio is called the regenerative braking ratio coefficient, which determines the proportion of motor braking and brake braking. Therefore, the regenerative braking proportional coefficient is used as the output variable of the fuzzy controller.
In recent years, research on energy recovery technology has mainly focused on two aspects: rule-based control and optimized control. However, the implementation of optimized control is complex and difficult to apply in practical applications. In contrast, rule-based control strategies have a simple logic and are relatively easy to implement [27]. Therefore, using a fuzzy control method, with road gradient (sl), braking intensity (z), and vehicle speed (v) as input variables, and the regenerative braking proportional coefficient (k) as the output variable, a regenerative braking fuzzy controller structure is designed as shown in Figure 4. braking to achieve the best energy recovery efficiency. This distribution ratio is called the regenerative braking ratio coefficient, which determines the proportion of motor braking and brake braking. Therefore, the regenerative braking proportional coefficient is used as the output variable of the fuzzy controller.
In recent years, research on energy recovery technology has mainly focused on two aspects: rule-based control and optimized control. However, the implementation of optimized control is complex and difficult to apply in practical applications. In contrast, rulebased control strategies have a simple logic and are relatively easy to implement [27]. Therefore, using a fuzzy control method, with road gradient (sl), braking intensity (z), and vehicle speed (v) as input variables, and the regenerative braking proportional coefficient (k) as the output variable, a regenerative braking fuzzy controller structure is designed as shown in Figure 4.  This paper uses expert knowledge and experience to define fuzzy sets and rules, and calibrates them through comparison with the actual situation. According to the calibration results, we obtain the fuzzy reasoning process, as shown in Figure 5 Table 2 for different road gradient thresholds (L, M, and H).  This paper uses expert knowledge and experience to define fuzzy sets and rules, and calibrates them through comparison with the actual situation. According to the calibration results, we obtain the fuzzy reasoning process, as shown in Figure 5 Table 2 for different road gradient thresholds (L, M, and H). braking to achieve the best energy recovery efficiency. This distribution ratio is called the regenerative braking ratio coefficient, which determines the proportion of motor braking and brake braking. Therefore, the regenerative braking proportional coefficient is used as the output variable of the fuzzy controller.
In recent years, research on energy recovery technology has mainly focused on two aspects: rule-based control and optimized control. However, the implementation of optimized control is complex and difficult to apply in practical applications. In contrast, rulebased control strategies have a simple logic and are relatively easy to implement [27]. Therefore, using a fuzzy control method, with road gradient (sl), braking intensity (z), and vehicle speed (v) as input variables, and the regenerative braking proportional coefficient (k) as the output variable, a regenerative braking fuzzy controller structure is designed as shown in Figure 4.  This paper uses expert knowledge and experience to define fuzzy sets and rules, and calibrates them through comparison with the actual situation. According to the calibration results, we obtain the fuzzy reasoning process, as shown in Figure 5 Table 2 for different road gradient thresholds (L, M, and H).

Simulink Simulation Validation
In order to verify the influence of the regenerative braking strategy based on a single pedal on the endurance of electric trucks, this paper inputs the NEDC (New European Driving Cycle) condition into the Matlab/Simulink platform simulation model. It compares it with the regenerative braking control strategy of the original vehicle using the look-up table method to verify the single-pedal regenerative braking control strategy. Figure 6 is the regenerative braking strategy simulation result based on a single pedal under NEDC conditions. The torque output by the motor can demonstrate the degree of regenerative braking. The charging and discharging current of the battery during the driving process of the car can more intuitively depict the energy recovery effect of regenerative braking. The SOC comparison of the battery can demonstrate the impact of the two control strategies on the endurance of the electric truck. Under NEDC conditions, the regenerative braking strategy based on a single pedal outputs more motor braking than the regenerative braking strategy based on the look-up table method, and the charging current of the power battery is also larger. The regenerative braking strategy based on a single pedal performs better in energy consumption than the regenerative braking strategy based on the look-up table method.
The comparison of energy recovery effects between the single-pedal-based regenerative braking strategy and the look-up-table-based regenerative braking strategy under the NEDC driving cycle is shown in Table 3. From Table 3, it can be observed that under the same total braking energy during the NEDC driving cycle, the single-pedal-based regenerative braking strategy recovers an additional 211 kJ of energy compared to the original vehicle's look-up-table-based regenerative braking strategy. The brake energy recovery rate of the proposed strategy in this study reaches 50.8%, while the original vehicle achieves a rate of only 30.5%, resulting in a 20.3% improvement. Compared to the original vehicle's effective braking energy recovery rate of 5.1%, the regenerative braking strategy proposed in this study improves it by 3.5% to achieve an 8.6% recovery rate. This indicates that the single-pedal-based regenerative braking strategy proposed in this study performs better regarding energy consumption for commercial vehicles with a single pedal. braking. The SOC comparison of the battery can demonstrate the impact of the two control strategies on the endurance of the electric truck. Under NEDC conditions, the regenerative braking strategy based on a single pedal outputs more motor braking than the regenerative braking strategy based on the look-up table method, and the charging current of the power battery is also larger. The regenerative braking strategy based on a single pedal performs better in energy consumption than the regenerative braking strategy based on the look-up table method. The comparison of energy recovery effects between the single-pedal-based regenerative braking strategy and the look-up-table-based regenerative braking strategy under the NEDC driving cycle is shown in Table 3. From Table 3, it can be observed that under the same total braking energy during the NEDC driving cycle, the single-pedal-based regenerative braking strategy recovers an additional 211 kJ of energy compared to the original vehicle's look-up-table-based regenerative braking strategy. The brake energy recovery rate of the proposed strategy in this study reaches 50.8%, while the original vehicle achieves a rate of only 30.5%, resulting in a 20.3% improvement. Compared to the original

Real Vehicle Verification with a Constant-Speed Car-Following Experiment
In order to verify the energy consumption optimization effect of the regenerative braking strategy on the actual vehicle and ensure that the control strategy is the only variable, a constant-speed car-following experiment plan is first designed. Then, under the premise of ensuring safety conditions, the control strategy proposed in this article is verified based on the actual vehicle.

Design of Constant-Speed Car-Following Experiment Scheme
Firstly, the vehicle control strategy model is generated into c code through the c code automatic generation technology of Simulink software (Matlab2021a, MathWorks, Natick, MA, USA). Then, the c code is put into Codewarrior software (V11.0, Freescale, Austin, TX, USA) and combined with the underlying software (V11.0, Freescale, Austin, TX, USA) of the vehicle controller. Finally, the vehicle control program is generated in an abs file format and burned into the vehicle controller.
The VCU software (V11.0, Freescale, Austin, TX, USA) burning process is shown in Figure 7. In the software burning process, the VCU vehicle controller needs an external 24 V low-voltage power supply to supply power to itself. It is connected to the positive and negative interfaces of the VCU vehicle controller A port with a pin. One end of the PE burner is connected to the USB interface of the computer, and the other is connected to the VCU vehicle controller. The software used by the PE burner is a function of the Codewarrior software. After the hardware is connected, it is necessary to configure the corresponding environment in the Codewarrior, select the corresponding burner hardware model and environmental variables, and then select the program file that needs to be burned. Click the run button and wait for the program to be burned.

Real Vehicle Verification with a Constant-Speed Car-Following Experiment
In order to verify the energy consumption optimization effect of the regenerative braking strategy on the actual vehicle and ensure that the control strategy is the only variable, a constant-speed car-following experiment plan is first designed. Then, under the premise of ensuring safety conditions, the control strategy proposed in this article is verified based on the actual vehicle.

Design of Constant-Speed Car-Following Experiment Scheme
Firstly, the vehicle control strategy model is generated into c code through the c code automatic generation technology of Simulink software (Matlab2021a, MathWorks, Natick, Massachusetts, USA). Then, the c code is put into Codewarrior software (V11.0, Freescale, Austin, Texas, USA) and combined with the underlying software (V11.0, Freescale, Austin, Texas, USA) of the vehicle controller. Finally, the vehicle control program is generated in an abs file format and burned into the vehicle controller.
The VCU software (V11.0, Freescale, Austin, Texas, USA) burning process is shown in Figure 7. In the software burning process, the VCU vehicle controller needs an external 24 V low-voltage power supply to supply power to itself. It is connected to the positive and negative interfaces of the VCU vehicle controller A port with a pin. One end of the PE burner is connected to the USB interface of the computer, and the other is connected to the VCU vehicle controller. The software used by the PE burner is a function of the Codewarrior software. After the hardware is connected, it is necessary to configure the corresponding environment in the Codewarrior, select the corresponding burner hardware model and environmental variables, and then select the program file that needs to be burned. Click the run button and wait for the program to be burned. Then, two experimental vehicles with similar battery states and the same model were used. One vehicle was equipped with the original vehicle control strategy, while the other adopted the single-pedal-based regenerative braking control strategy proposed in this study. Both vehicles were fully loaded, fully charged, and followed each other in a carfollowing scenario. During the car-following experiment, the speeds of the two vehicles were synchronized, and the drivers' usage of the accelerator pedal was kept similar. The advantage of experimenting with a constant-speed car-following scenario is controlling Then, two experimental vehicles with similar battery states and the same model were used. One vehicle was equipped with the original vehicle control strategy, while the other adopted the single-pedal-based regenerative braking control strategy proposed in this study. Both vehicles were fully loaded, fully charged, and followed each other in a carfollowing scenario. During the car-following experiment, the speeds of the two vehicles were synchronized, and the drivers' usage of the accelerator pedal was kept similar. The advantage of experimenting with a constant-speed car-following scenario is controlling variables and minimizing driver style's influence on the experiment. The experimental process is illustrated in Figure 8. variables and minimizing driver style's influence on the experiment. The experimental process is illustrated in Figure 8. In the experimental process, after loading the vehicles with cargo, the vehicles were first weighed using a weighing scale to ensure that the overall vehicle mass of both cars was the same. The weights of the two vehicles were adjusted to be equal without the drivers getting out of the vehicles. During the car-following experiment, real-time data from both vehicles were recorded, including displayed mileage and remaining state of charge (SOC). The displayed SOC was only used as a reference value to obtain accurate energy consumption data. In contrast, the energy charged into the power battery until it reached total capacity was considered the final result value. This approach eliminated the influence of SOC value discrepancies on the experiment.
Additionally, after completing the tests, both vehicles were simultaneously charged on a charging station to eliminate the impact of charging station efficiency on the experimental results. The amount of energy charged was then used as the result. To eliminate randomness, multiple experiments were conducted.
The hardware connections for data acquisition are illustrated in Figure 9. The experimental data are essential for analyzing the experimental results. This study used the INCA and CAN box tools to collect data from the actual vehicle. The INCA and CAN box tools interact with the vehicle through the CAN bus and require a connection to the vehicle's CAN system. One end of the INCA tool is connected to the computer via a In the experimental process, after loading the vehicles with cargo, the vehicles were first weighed using a weighing scale to ensure that the overall vehicle mass of both cars was the same. The weights of the two vehicles were adjusted to be equal without the drivers getting out of the vehicles. During the car-following experiment, real-time data from both vehicles were recorded, including displayed mileage and remaining state of charge (SOC). The displayed SOC was only used as a reference value to obtain accurate energy consumption data. In contrast, the energy charged into the power battery until it reached total capacity was considered the final result value. This approach eliminated the influence of SOC value discrepancies on the experiment.
Additionally, after completing the tests, both vehicles were simultaneously charged on a charging station to eliminate the impact of charging station efficiency on the experimental results. The amount of energy charged was then used as the result. To eliminate randomness, multiple experiments were conducted.
The hardware connections for data acquisition are illustrated in Figure 9.
World Electr. Veh. J. 2021, 12, x FOR PEER REVIEW 11 of 16 variables and minimizing driver style's influence on the experiment. The experimental process is illustrated in Figure 8. In the experimental process, after loading the vehicles with cargo, the vehicles were first weighed using a weighing scale to ensure that the overall vehicle mass of both cars was the same. The weights of the two vehicles were adjusted to be equal without the drivers getting out of the vehicles. During the car-following experiment, real-time data from both vehicles were recorded, including displayed mileage and remaining state of charge (SOC). The displayed SOC was only used as a reference value to obtain accurate energy consumption data. In contrast, the energy charged into the power battery until it reached total capacity was considered the final result value. This approach eliminated the influence of SOC value discrepancies on the experiment.
Additionally, after completing the tests, both vehicles were simultaneously charged on a charging station to eliminate the impact of charging station efficiency on the experimental results. The amount of energy charged was then used as the result. To eliminate randomness, multiple experiments were conducted.
The hardware connections for data acquisition are illustrated in Figure 9. The experimental data are essential for analyzing the experimental results. This study used the INCA and CAN box tools to collect data from the actual vehicle. The INCA and CAN box tools interact with the vehicle through the CAN bus and require a connection to the vehicle's CAN system. One end of the INCA tool is connected to the computer via a The experimental data are essential for analyzing the experimental results. This study used the INCA and CAN box tools to collect data from the actual vehicle. The INCA and CAN box tools interact with the vehicle through the CAN bus and require a connection to the vehicle's CAN system. One end of the INCA tool is connected to the computer via a USB interface. In contrast, the other end is connected to the vehicle's pure electric mini truck CAN system using a DB9 connector. The vehicle data are transmitted to the computer through the INCA tool after being processed via the vehicle's CAN communication lines. The computer, equipped with the INCA software (7.2, ETAS, Shanghai, China), can perform real-time monitoring, data acquisition, and calibration tasks on the vehicle. Additionally, the CAN box can be connected for message acquisition, and the communication transmission of the CAN box can be achieved through calibrated CAN.

Analysis of the Constant-Speed Car-Following Experiment Results
To verify the torque-following performance of the single-pedal system, a segment of driving data were collected while the driver did not engage the emergency braking pedal. The experimental results of the effectiveness verification of the single-pedal control strategy on the actual vehicle are shown in Figure 10. To verify the torque-following performance of the single-pedal system, a segment of driving data were collected while the driver did not engage the emergency braking pedal. The experimental results of the effectiveness verification of the single-pedal control strategy on the actual vehicle are shown in Figure 10.  Figure 10 shows that during the first 0 to 5 s, when the driver does not operate the pedal, the torque output is 0, and the vehicle speed is 0. From the 5th to the 8th second, when the driver presses the pedal in single-pedal mode, the motor torque follows the pedal aperture. It starts to output positive torque, increasing vehicle speed. From the 128th to the 135th s, there is a sharp decrease in the pedal aperture, and the motor starts to output negative torque, enabling regenerative braking and causing the vehicle to decelerate. The motor's positive torque output generally follows the pedal aperture trend and can generate negative torque for regenerative braking as the pedal aperture decreases.  Figure 10 shows that during the first 0 to 5 s, when the driver does not operate the pedal, the torque output is 0, and the vehicle speed is 0. From the 5th to the 8th second, when the driver presses the pedal in single-pedal mode, the motor torque follows the pedal aperture. It starts to output positive torque, increasing vehicle speed. From the 128th to the 135th s, there is a sharp decrease in the pedal aperture, and the motor starts to output negative torque, enabling regenerative braking and causing the vehicle to decelerate. The motor's positive torque output generally follows the pedal aperture trend and can generate negative torque for regenerative braking as the pedal aperture decreases.
The energy consumption test results are shown in Figure 11. Figure 11a displays the vehicle speed data throughout the entire duration of the test, while Figure 11b represents the state of charge (SOC) data for the same test. The test lasted for over three hours. Figure 11b shows that using the single-pedal regenerative braking strategy, the electric delivery truck had a remaining SOC of 18%, which is 6% higher than the remaining SOC of 12% achieved with the original vehicle control strategy. The energy consumption test results are shown in Figure 11. Figure 11a displays the vehicle speed data throughout the entire duration of the test, while Figure 11b represents the state of charge (SOC) data for the same test. The test lasted for over three hours. Figure  11b shows that using the single-pedal regenerative braking strategy, the electric delivery truck had a remaining SOC of 18%, which is 6% higher than the remaining SOC of 12% achieved with the original vehicle control strategy. In order to eliminate the influence of randomness on the vehicle experiments and consider various factors such as experimental conditions, this study conducted an additional four real-world experiments on regenerative braking control strategy based on constant-speed car-following. Table 4 presents the results of the five energy consumption experiments using the single-pedal regenerative braking control strategy.  In order to eliminate the influence of randomness on the vehicle experiments and consider various factors such as experimental conditions, this study conducted an additional four real-world experiments on regenerative braking control strategy based on constant-speed car-following. Table 4 presents the results of the five energy consumption experiments using the single-pedal regenerative braking control strategy.
The experimental results in Table 4 show that in the five energy consumption tests, the single-pedal regenerative braking strategy proposed in this study outperformed the original vehicle strategy regarding energy consumption for the electric delivery truck. Among them, the second test showed the smallest difference, with a remaining SOC value of 20% and a rechargeable energy of 78.8 kWh for the proposed strategy, compared to a remaining SOC value of 15% and a rechargeable energy of 81.6 kWh for the original vehicle strategy. The proposed strategy achieved a 3.45% optimization compared to the original vehicle strategy. The third test showed the largest difference, with a remaining SOC value of 21% and a rechargeable energy of 77.8 kWh for the proposed strategy, compared to a remaining SOC value of 13% and a rechargeable energy of 82.6 kWh for the original vehicle strategy, resulting in a 5.81% optimization. The differences in the test results fall within the controllable error range of the real-world experiments, indicating the effectiveness of the energy consumption tests.
In summary, the single-pedal regenerative braking strategy proposed in this study achieved a maximum optimization rate of 5.81% and an average optimization rate of 4.33% compared to the original vehicle strategy. The proposed regenerative braking strategy demonstrates better fuel efficiency.

Discussion
Like the existing references [23][24][25][26], the single-pedal regenerative braking strategy proposed in this paper is verified in the simulation, and the braking energy recovery rate and effective braking energy recovery rate are increased by 20.3% and 3.5%, respectively. However, the simulation cannot fully simulate the actual road environment factors. The real-vehicle verification can take into account the system integration, provide more accurate and real data, and evaluate the performance, safety and reliability of the strategy. Therefore, this paper carried out a constant-speed car-following experiment. The results show that the maximum optimization rate of energy consumption of the single-pedal regenerative braking control strategy proposed in this paper reaches 5.81%, and the average optimization rate reaches 4.33%, which can effectively reduce energy consumption and improve the economy of single-pedal pure electric commercial vehicles.
The fuzzy control method used in this paper is based on experience and has high requirements for researchers. Therefore, future research can explore the combination of neural networks and swarm intelligence algorithms to further improve the performance of the single-pedal regenerative braking control strategy. By combining the learning ability of neural networks with the optimization ability of swarm intelligence algorithms, a more intelligent and adaptive regenerative braking control strategy can be realized. This will bring greater improvements to the energy recovery and driving experience of the vehicle's braking system.

Conclusions
The single-pedal regenerative braking strategy proposed in this study divides the single-pedal braking system into two modes: pure motor braking and hybrid braking. Fuzzy control is utilized to allocate the motor braking ratio during hybrid braking. Simulation results demonstrate that the proposed strategy's braking energy recovery and effective braking energy recovery rates are improved by 20.3% and 3.5%, respectively. Real-world car-following experiments show that the maximum optimization rate of the proposed strategy for energy consumption reaches 5.81%, with an average optimization rate of 4.33%.
The proposed strategy effectively improves energy utilization and increases the driving range of electric vehicles while reducing driver burden.

Conflicts of Interest:
The authors declare no conflict of interest.

Abbreviations
The following abbreviations are used in this manuscript: v The value of velocity v max The maximum threshold of velocity m 1 , m 2 The characteristic coefficient pel The value of pedal aperture pel am , pel cm , and pel bm The maximum value of the pedal aperture corresponding to the acceleration interval, the cruise interval, and the deceleration interval, respectively pel The value of the single-pedal aperture pel emer The emergency braking pedal aperture, SOC The battery state of charge x pel The difference between the current pedal aperture and the previous pedal aperture x pel_max The given threshold of x pel v pel The displacement change rate of the pedal aperture v pel_max The given threshold of v pel sl The road gradient sl max The given threshold of sl T m The current regenerative braking torque α The road gradient conversion coefficient β The pedal displacement rate conversion coefficient γ The pedal displacement conversion coefficient, The threshold pedal displacement for increasing the brake pedal displacement T m−max The peak torque of the motor. z Braking intensity k The regenerative braking proportional coefficient