Regenerative braking cooperative control method based on real-time slope recognition and brake temperature estimation

Abstract

The application relates to the technical field of vehicle brake control, and particularly discloses a regenerative brake cooperative control method based on real-time slope identification and brake temperature estimation. The method is used to solve the problems that in the prior art, slope information is used for qualitative judgment, brake temperature management relies on after-compensation, and brake force distribution lacks multivariable cooperative optimization. The method comprises the following steps: in a no-acceleration instruction state, a road slope angle is iteratively calculated based on longitudinal acceleration and vehicle speed by using a vehicle longitudinal dynamics equation; a total brake torque is calculated based on the slope angle, a preset target deceleration and the vehicle speed; a mechanical brake temperature is real-time recursively estimated based on a lumped parameter thermal model; a mechanical brake torque proportion coefficient is output by a fuzzy controller with the slope angle, the vehicle speed, a battery state of charge and the estimated temperature as input variables, and the coefficient is forcibly limited to not more than 0.3 when the estimated temperature reaches 200 DEG C; regenerative brake torque and mechanical brake torque are calculated and distributed according to the coefficient and the total brake torque, and the mechanical brake torque is used for compensation when the regenerative capacity is insufficient, so that the total brake torque can be accurately matched with the target deceleration.

Description

Technical Field

[0001] This application relates to the field of automotive braking control technology, and more specifically to a regenerative braking coordinated control method based on real-time slope identification and brake temperature prediction. Background Technology

[0002] Regenerative braking technology is widely used in electric vehicles. By switching the motor to generator mode during deceleration or downhill driving, the vehicle's kinetic or gravitational potential energy is converted into electrical energy to recharge the battery, thereby improving range. For shared electric vehicles, due to high usage frequency and long charging intervals, regenerative braking is crucial for extending the driving range per charge. In existing technologies, some solutions (such as CN107458531A) have incorporated a slope sensing module, but they only use slope information as a binary "uphill / downhill" switch signal to trigger or disable regenerative braking, without incorporating the slope angle as a continuous quantitative parameter in the physical calculation of the total braking torque. In fact, the component of gravity along the slope direction during downhill driving is mg·sinα. For every 1° increase in slope, the required braking torque increases by approximately 2%-3%. Ignoring this quantitative relationship will lead to a mismatch between braking force supply and actual demand, resulting in either increased braking distance or a significant decrease in energy recovery efficiency.

[0003] Furthermore, while existing technologies (such as CN115771497B) employ brake temperature detection to determine the state of thermal fade and compensate by increasing regenerative braking force or mechanical braking pressure after the temperature rises, this method belongs to the logic of "post-event compensation"—that is, measures are taken only after the friction coefficient has decreased due to high temperature and the braking performance has actually been damaged, which cannot eliminate the safety hazards that have already occurred. Other solutions (such as CN118404998B) use thermodynamic models to predict brake temperature, but their prediction results are only used to trigger external cooling devices (such as fans or coolant pumps), and temperature is not used as an intrinsic constraint on the braking force distribution strategy, thus failing to reduce heat generation at the source.

[0004] In terms of the distribution strategy between regenerative braking force and mechanical braking force, existing control schemes generally suffer from simplistic and static problems: they either use a fixed ratio (such as 7:3) or adjust the distribution weight only based on the battery SOC, lacking a coordinated response to multiple variables such as slope, vehicle speed, load, and temperature; or they only activate regenerative braking when the brake pedal is pressed, ignoring the energy recovery potential during coasting downhill. More importantly, existing technologies are mostly designed for four-wheeled vehicles or hybrid vehicles, relying on brake pressure sensors, wheel speed sensor arrays, multi-channel temperature sensors, and high-performance ECUs. However, shared electric bicycles are limited by cost and space, and cannot directly reuse such high-hardware configurations, resulting in poor adaptability, low accuracy, and insufficient safety in their control strategies during actual deployment.

[0005] In summary, existing technologies remain at the qualitative judgment level in utilizing slope information, without achieving quantitative modeling; they rely on post-event compensation or external cooling for brake temperature management, lacking a pre-event constraint mechanism; and they lack multi-dimensional dynamic collaborative optimization in braking force distribution, and are highly dependent on hardware, making it difficult to adapt to the low-cost, lightweight platform of shared electric bicycles. This results in systemic defects such as low energy recovery efficiency, significant braking safety hazards, and accelerated mechanical brake wear when operating in areas with frequent slopes, such as mountains and hills. Summary of the Invention

[0006] To address the shortcomings of existing technologies, this application provides a regenerative braking coordinated control method based on real-time ramp identification and brake temperature prediction, comprising:

[0007] S1. In the absence of acceleration command, based on the acquired longitudinal acceleration a_x and vehicle speed v, the road slope angle α is obtained through iterative calculation using the vehicle longitudinal dynamics equation. The expression of the vehicle longitudinal dynamics equation is: sinα = (m·a_x - μ·m·g·cosα - 0.5·ρ·Cd·A·v²) / (m·g), where m is the vehicle mass, μ is the rolling resistance coefficient, ρ is the air density, Cd is the wind resistance coefficient, A is the vehicle frontal area, and the vehicle mass m is obtained by inverse calculation of air resistance and rolling resistance in the absence of acceleration command.

[0008] S2. Based on the slope angle α, the preset target deceleration a_target, and the vehicle speed v, calculate the total braking torque T_total required to maintain the preset target deceleration using vehicle dynamics formulas. The formula for calculating the total braking torque T_total is: T_total = [m·|a_target| + m·g·sinα + μ·m·g·cosα + 0.5·ρ·Cd·A·v²] · r, where the preset target deceleration a_target ranges from 0.8 m / s² to 2.0 m / s², r is the wheel rolling radius, and when α≥0, m·g·sinα is added as a positive braking force component, and when α≤0, m·g·sinα is subtracted as a negative braking force component.

[0009] S3. Based on a lumped-parameter thermal model of mechanical braking power, heat dissipation parameters, and ambient temperature, the mechanical brake temperature T_brake is recursively estimated in real time. S4. Using slope angle α, vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake as input variables, a fuzzy controller outputs the mechanical braking torque ratio coefficient K. When the estimated temperature T_brake ≥ 200℃, K is forcibly limited to ≤ 0.3. The input variables of the fuzzy controller include slope angle α, vehicle speed v, battery SOC, and estimated temperature T_brake. The corresponding fuzzy rules of the fuzzy controller... The core rules of the system are as follows: When the slope angle α is a steep downhill slope (PB), the vehicle speed v is high (H), the battery state of charge (SOC) is low (L), and the estimated temperature T_brake is normal (N), the mechanical braking torque ratio coefficient K is output as very small (VS); when the estimated temperature T_brake is high (P), regardless of the slope angle α, vehicle speed v, or battery state of charge (SOC), the mechanical braking torque ratio coefficient K is output as no more than 0.3; when the slope angle α is a steep uphill slope (NB), regardless of the vehicle speed v, battery state of charge (SOC), or estimated temperature T_brake, the mechanical braking torque ratio coefficient K is output as very large (VL).

[0010] S5. Calculate the regenerative braking torque T_reg = min(T_reg_max, (1-K)·T_total) based on the mechanical braking torque ratio coefficient K and the total braking torque T_total, and calculate the mechanical braking torque T_mech = K·T_total; the upper limit of the regenerative braking torque T_reg_max is determined by T_reg_max = min(T_motor_max(ω), P_chg_max(SOC, T_bat) / ω), where T_motor_max(ω) is obtained by looking up the motor generating torque MAP table pre-stored in the controller's non-volatile memory, P_chg_max is sent in real time by the battery management system BMS via the CAN bus, and T_bat is the battery temperature; when T_reg_max is less than (1-K)·T_total, the mechanical braking torque T_mech is compensated to T_mech = T_total - T_reg to ensure that the total braking torque accurately matches the target deceleration.

[0011] Optionally, in one implementation of the technical solution of this application, the real-time recursive estimation of the mechanical brake temperature T_brake specifically includes:

[0012] Within each sampling period Ts, based on the mechanical braking torque T_mech(k-1), wheel speed ω(k-1), ambient temperature T_ambient, and thermal parameters h, A, c, and m_brake from the previous period, the following is executed: T_brake(k) = T_brake(k-1) + [T_mech(k-1)·ω(k-1) - h·A·(T_brake(k-1) - T_ambient)] / (c·m_brake) · Ts, where the sampling period Ts ≤ 200ms;

[0013] The thermal parameters in the lumped-parameter thermal model are: m_brake = 0.05 kg, c = 500 J / (kg·K), A = 0.002 m², h = 30 W / (m²·K). T_ambient is collected by the vehicle's ambient temperature sensor, and T_brake is initialized to the current value of T_ambient when the vehicle is powered on and started.

[0014] Optionally, in one implementation of the technical solution of this application, the fuzzy universe of discourse for the slope angle α is [-10°, 10°], divided into five fuzzy sets: large uphill (NB, center point 8°), small uphill (NS, center point 3°), flat road (ZO, center point 0°), small downhill (PS, center point 3°), and large downhill (PB, center point 8°), with the membership function being a trigonometric function; the fuzzy universe of discourse for the vehicle speed v is [0, 30] km / h, divided into three fuzzy sets: low speed (L, center point 5 km / h), medium speed (M, center point 15 km / h), and high speed (H, center point 25 km / h). The fuzzy universe of discourse for the battery's state of charge (SOC) is [0, 100]%, divided into three fuzzy sets: low (L, center point 20%), medium (M, center point 60%), and high (H, center point 90%), with triangular membership functions. The fuzzy universe of discourse for the estimated temperature (T_brake) is [25℃, 300℃], divided into three fuzzy sets: normal (N, center point 60℃), slightly high (P, center point 160℃), and dangerous (D, center point 250℃), with triangular membership functions. When T_brake ≥ 200℃, K ≤ 0.3 is forcibly restricted, and it does not participate in fuzzy inference.

[0015] Optionally, in one implementation of the technical solution of this application, the fuzzy universe of the output variable K is [0, 1], which is divided into five fuzzy sets: very small (VS, center point 0.05), small (S, center point 0.2), medium (M, center point 0.4), large (L, center point 0.6), and very large (VL, center point 0.85), and the centroid method is used for defuzzification.

[0016] Optionally, in one implementation of the technical solution of this application, in the fuzzy rule base, when the estimated temperature T_brake is in the range of 120℃ to 200℃, the K value is prohibited from increasing, the K value of the previous cycle remains unchanged, and only the K value is allowed to decrease or remain unchanged; when the estimated temperature T_brake rises from below 120℃ to 120℃, the K value locking mechanism is triggered, the K value is prohibited from increasing, and the locking is not released until T_brake drops below 100℃.

[0017] Optionally, in one implementation of the technical solution of this application, the method is applicable to shared electric bicycles, and its drive system is a hub-type brushless DC motor.

[0018] Compared with the prior art, the technical solution of this application has the following beneficial effects:

[0019] By continuously estimating the road slope angle α using accelerometers and wheel speed sensors without acceleration commands, and then accurately calculating the total braking torque required to maintain stable vehicle deceleration based on this slope angle α in conjunction with the target deceleration and vehicle speed, the braking force demand directly responds to the physical characteristics of the road slope, overcoming the problem of inaccurate braking force distribution caused by relying solely on experience or qualitative judgment in traditional methods. Simultaneously, the mechanical brake temperature is predicted in real-time using a lumped-parameter thermal model. This temperature is used as a key feedback variable, inputting along with the slope angle α, vehicle speed v, and battery state of charge (SOC) into a fuzzy controller, dynamically outputting the mechanical braking torque proportion coefficient K. Furthermore, when the estimated temperature is greater than or equal to 200℃, K is forcibly limited to be less than or equal to 0.3, thereby achieving proactive constraint on the risk of brake thermal fade. Finally, based on K and the total braking torque, the regenerative braking torque and mechanical braking torque are coordinated and allocated to form a dynamic collaborative control logic driven by quantitative slope input, with pre-existing temperature constraints as the safety boundary, and integrating multiple operating condition parameters. This effectively solves the problem of multi-factor collaborative optimization of regenerative braking and mechanical braking in existing technologies, which lacks quantitative slope response and forward-looking thermal constraints, and achieves the comprehensive effect of improving energy recovery efficiency, preventing brake thermal fade, and ensuring the safe and stable operation of the braking system. Attached Figure Description

[0020] Figure 1 This is a flowchart illustrating a regenerative braking coordinated control method based on real-time slope identification and brake temperature prediction in an embodiment of this application. Detailed Implementation

[0021] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0022] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this invention are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of the invention described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover a non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0023] As described in the background section, existing regenerative braking systems for shared electric vehicles only treat slope information as a binary "uphill / downhill" switch signal in slope scenarios, without using it as a continuous quantitative parameter in braking torque calculation. At the same time, brake temperature management relies on passive compensation or external cooling triggering after thermal fade occurs. The distribution strategy of regenerative braking force and mechanical braking force lacks coordinated response to multiple variables such as slope, temperature, vehicle speed, and SOC. This results in a mismatch between braking force supply and actual demand, difficulty in preventing thermal fade risk, and low energy recovery efficiency when operating in areas with frequent downhill slopes such as mountains and hills. Furthermore, existing solutions with high hardware dependence cannot be adapted to the low-cost, lightweight platform of shared electric bicycles.

[0024] After long-term research, the inventors discovered that the most fundamental reason for the aforementioned predicament lies in the fact that existing technologies incorrectly separate the two key dynamic variables, "slope" and "temperature," from the core logic of braking force allocation decisions: slope is reduced to a simple enabling condition rather than a physical input driving braking force demand; temperature is treated as an externally modifiable environmental variable rather than an intrinsic constraint directly affecting the proportion of braking force. This prevents the system from establishing a real-time, closed-loop, physically driven dynamic balance among "braking force demand—regenerative capacity—thermal safety," instead employing fragmented, lagging, and static control logic across multiple dimensions—essentially a low-order control mindset of "step-by-step response, each acting independently." A deeper contradiction lies in the fact that without quantitative modeling of slope, the system cannot accurately calculate the actual braking force demand; and without considering temperature as a pre-emptive constraint, it cannot proactively suppress heat sources before thermal risks materialize.

[0025] Based on the above findings, the inventors proposed a design that uses both "continuous quantitative estimation of the slope angle" and "predictive constraint of brake temperature" as direct input variables for braking force distribution decisions, constructing a collaborative control architecture with "total braking torque demand - regenerative capacity upper limit - temperature safety threshold" as triangular constraints. The core of this design is that instead of using the slope as a "gate signal" for activating regenerative braking, it is used as the sole physical input parameter for calculating the total braking torque T_total, ensuring precise matching of braking force supply to changes in the gravitational component. Simultaneously, instead of waiting for brake temperature to exceed the limit before compensation, the temperature T_brake is predicted in real-time using a thermal model, and before reaching the critical point of 200℃, the mechanical braking force ratio coefficient K is forcibly limited to within 0.3, proactively curbing the temperature rise at its source. This concept is the first to integrate two previously isolated physical quantities, "gradient quantification" and "temperature pre-constraint," into the decision-making closed loop of braking force distribution. By using a fuzzy controller with K as the mediator, the composition ratio of regeneration and mechanical braking force is dynamically adjusted. Thus, without adding any hardware, a fundamental leap is achieved from "passive response" to "active constraint" and from "single-variable adjustment" to "multi-physical quantity synergy."

[0026] Guided by this ingenious concept, the inventors proposed a regenerative braking collaborative control method based on real-time slope identification and brake temperature prediction.

[0027] Example 1: As Figure 1 As shown in the embodiments of this application, the regenerative braking coordinated control method based on real-time slope identification and brake temperature prediction includes:

[0028] S1. In the absence of an acceleration command, based on the acquired longitudinal acceleration a_x and vehicle speed v, the road slope angle α is obtained through iterative calculation using the vehicle longitudinal dynamics equation.

[0029] In this embodiment, the expression for the vehicle's longitudinal dynamics equation is: sinα = (m·a_x - μ·m·g·cosα - 0.5·ρ·Cd·A·v²) / (m·g), where m is the vehicle mass, μ is the rolling resistance coefficient, ρ is the air density, Cd is the drag coefficient, A is the vehicle's frontal area, and the vehicle mass m is obtained by inverse calculation of air resistance and rolling resistance under no acceleration command conditions.

[0030] Specifically, the estimation of the slope angle α utilizes the vehicle's existing triaxial accelerometer and wheel speed sensors. The accelerometer is used to collect the vehicle's longitudinal acceleration a_x, and its installation direction is consistent with the vehicle's forward direction. The output value is the acceleration component of the vehicle in the traveling direction. The wheel speed sensors are installed at the wheel hubs to collect wheel speed signals, which are then converted into vehicle speed v by combining the wheel rolling radius r. In the absence of acceleration commands, i.e., when the electronic throttle opening is zero and the motor does not output driving force, the vehicle is only affected by gravity, rolling resistance, air resistance, and braking force. Based on the dynamic relationship under this condition, the value of the slope angle α is derived through a mathematical model. This estimation process does not rely on a preset slope threshold or binary judgment, but continuously outputs a continuous slope angle α that changes with the vehicle's operating state, achieving real-time quantitative identification of the road slope.

[0031] At this point, the longitudinal dynamic equation of the vehicle can be expressed as: m·a_x = m·g·sinα - μ·m·g·cosα - 0.5·ρ·Cd·A·v², where the left side of the equation represents the mass inertial force, and the right side represents the gravity component along the slope, rolling resistance, and air resistance, respectively. Rearranging the equation, we obtain the expression for sinα: sinα = (m·a_x - μ·m·g·cosα - 0.5·ρ·Cd·A·v²) / (m·g). Since this equation contains a cosα term, α cannot be directly solved. Therefore, an iterative calculation method, such as the Newton-Raphson iteration method, is used to solve the nonlinear equation through successive approximations. In each iteration, the current estimated value of α is substituted into the right side of the equation, the residual is calculated, and the value of α is corrected based on the residual until the residual is less than a preset convergence threshold (e.g., 10⁻). 4 radian).

[0032] Furthermore, specifically, the vehicle mass *m* is calculated through inverse calculation of air resistance and rolling resistance under conditions of no acceleration command. Specifically, under conditions of no acceleration command, flat terrain, no wind, and the vehicle traveling at a constant speed *v0*, the motor current *I0* is collected. A dynamic balance equation is established by combining the motor torque constant *Kt*, motor efficiency *η*, air resistance coefficient *0.5·ρ·Cd·A·v0²*, and rolling resistance *μ·m·g*. The vehicle mass *m* = (η·Kt·I0 - 0.5·ρ·Cd·A·v0²) / (μ·g), and the average value of the calculation results is taken over three consecutive sampling periods to filter out instantaneous disturbances, so as to obtain a stable and reliable vehicle mass estimate. The m value obtained by this dynamic identification is directly applied to the estimation of the slope angle α and the calculation of the total braking torque T_total. This allows the slope identification and braking force demand derivation to get rid of the limitations of the fixed mass assumption, significantly improving the estimation accuracy under dynamic load change scenarios. This provides an accurate input basis for subsequent fuzzy collaborative control based on slope, vehicle speed, battery SOC and brake temperature. Especially under the constraint strategy of forcibly limiting the mechanical braking force ratio to ≤0.3 when the brake temperature exceeds 200℃, it ensures that the distribution of regenerative braking and mechanical braking is more accurate and safer, and finally achieves a comprehensive improvement in the collaborative control accuracy and reliability of the regenerative braking system under complex working conditions.

[0033] S2. Based on the slope angle α, the preset target deceleration a_target and the vehicle speed v, calculate the total braking torque T_tota required to maintain the preset target deceleration using vehicle dynamics formulas.

[0034] In this embodiment, based on the current road slope angle α, the preset target deceleration, and the vehicle's real-time speed, the total braking torque T_total required to achieve the target deceleration is calculated according to the basic physical relationships of vehicle longitudinal dynamics. This calculation process directly relies on the slope angle α as a continuous quantitative input, combined with the mass inertial drag component corresponding to the target deceleration, and the rolling resistance and air resistance components determined by the vehicle speed. These are unified into a single torque output value through dynamic equations, ensuring that the magnitude of the braking torque completely corresponds to the physical requirements for maintaining the target deceleration state under the current slope and speed conditions, without introducing any other unrecorded factors as adjustment criteria.

[0035] Furthermore, in one specific implementation, the formula for calculating the total braking torque T_total is: T_total = [m·|a_target| + m·g·sinα + μ·m·g·cosα + 0.5·ρ·Cd·A·v²] · r, where the preset target deceleration a_target ranges from 0.8 m / s² to 2.0 m / s², r is the wheel rolling radius, and when α≥0, m·g·sinα is added as a positive braking force component, and when α≤0, m·g·sinα is subtracted as a negative braking force component.

[0036] In the above implementation, by limiting the preset target deceleration a_target to a reasonable range of 0.8 m / s² to 2.0 m / s², and explicitly introducing the rolling resistance term μ·m·g·cosα and the air resistance term 0.5·ρ·Cd·A·v² into the calculation formula of the total braking torque T_total, and dynamically adjusting the superposition or cancellation logic of the gravity component m·g·sinα according to the positive or negative of the slope angle α, the calculation of the braking torque truly reflects the longitudinal dynamic requirements of the vehicle under actual slope conditions. This avoids insufficient or excessive braking force supply caused by ignoring air and rolling resistance or misjudging the direction of the gravity component, ensuring that the coordinated distribution of regenerative braking and mechanical braking is always based on an accurate physical model, improving energy recovery efficiency and ensuring the stability and safety of braking response, and ultimately achieving precise control of braking performance under the coupled effects of slope changes and multiple resistance factors.

[0037] S3. Based on the lumped parameter thermal model of mechanical braking power, heat dissipation parameters and ambient temperature, the temperature of the mechanical brake T_brake is predicted in real time using recursion.

[0038] In this embodiment, the temperature of the mechanical brake is recursively predicted in real time using a lumped-parameter thermal model that integrates mechanical braking power, heat dissipation parameters, and ambient temperature. Mechanical braking power is determined by the heat generated during brake operation, obtained through the product of braking torque and wheel speed, serving as the heat source input. Heat dissipation parameters characterize the brake's ability to dissipate heat to the environment, their magnitude determined by the material's thermal conductivity and convection characteristics. Ambient temperature serves as a reference value for heat exchange, used to calculate the temperature difference between the brake and the surrounding medium driving the heat dissipation process. Based on these three factors, the system establishes a thermal balance equation and, using a discrete recursive approach, progressively updates and outputs the current brake temperature prediction based on the temperature state at the previous moment and the current power input and heat dissipation rate, achieving dynamic tracking of temperature change trends without direct measurement.

[0039] S4. Using the slope angle α, vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake as input variables, the mechanical braking torque ratio coefficient K is output through a fuzzy controller. When the estimated temperature T_brake ≥ 200℃, K is forcibly limited to ≤ 0.3.

[0040] In this embodiment, the slope angle α, vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake are used as input variables. The mechanical braking torque ratio coefficient K is output by the fuzzy controller, so that the braking force distribution decision is dynamically adjusted based on these four real-time variables. When the estimated temperature T_brake reaches or exceeds 200°C, the system forcibly limits the value of K to no more than 0.3, thereby embedding a temperature threshold hard constraint in the fuzzy control logic, directly limiting the upper limit of the proportion of mechanical braking, and avoiding the continuous increase of the proportion of mechanical braking in the braking force distribution due to excessive temperature.

[0041] Furthermore, in one specific implementation, the input variables of the fuzzy controller include slope angle α, vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake. The core rule of the fuzzy rule base corresponding to the fuzzy controller is as follows: when the slope angle α is a steep downhill slope (PB), the vehicle speed v is high (H), the battery state of charge (SOC) is low (L), and the estimated temperature T_brake is normal (N), the mechanical braking torque ratio coefficient K is output as very small (VS); when the estimated temperature T_brake is high (P), regardless of the values ​​of slope angle α, vehicle speed v, and battery state of charge (SOC), the mechanical braking torque ratio coefficient K is output as no more than 0.3; when the slope angle α is a steep uphill slope (NB), regardless of the values ​​of vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake, the mechanical braking torque ratio coefficient K is output as very large (VL).

[0042] In this specific implementation, by setting the core rule that "when α is a steep downhill slope and the vehicle speed is high, the battery state of charge is low, and the brake temperature is normal, the output of K is very small," the system actively suppresses mechanical braking and maximizes regenerative energy recovery under the condition of high-speed downhill slope and acceptable battery charging. Simultaneously, by using the hard rule that "when α is a steep uphill slope, the output is large regardless of other variables," the system ensures that mechanical braking always dominates under uphill conditions, preventing control failure due to insufficient power or reverse energy flow in regenerative braking. Furthermore, by setting the rule that "when the brake temperature is high, K output is large regardless of the values ​​of other parameters," the system ensures that mechanical braking always dominates under uphill conditions, preventing control failure due to insufficient power or reverse energy flow in regenerative braking. The global constraint of "not exceeding 0.3" forms a dual guarantee with the temperature threshold forced limit, effectively avoiding the fluctuation of mechanical braking ratio caused by the uncertainty of fuzzy inference under high temperature conditions. This enables a multi-dimensional and accurate response to slope changes, vehicle speed dynamics, battery status and thermal safety boundaries. It solves the technical defects of existing technologies, such as low regeneration efficiency, unreasonable uphill braking force distribution and control instability at high temperatures, caused by coarse fuzzy control rules and lack of scenario-based priority. It achieves a comprehensive effect of significantly improved energy recovery rate, reliable protection of braking safety boundaries and enhanced system response predictability.

[0043] S5. Calculate the regenerative braking torque T_reg = min(T_reg_max, (1-K)·T_total) based on the mechanical braking torque ratio coefficient K and the total braking torque T_total, and calculate the mechanical braking torque T_mech = K·T_total.

[0044] In this embodiment, the mechanical braking torque T_mech is directly determined based on the product relationship between the mechanical braking torque proportion coefficient K and the total braking torque T_total. Simultaneously, the regenerative braking torque T_reg is determined based on the smaller of the upper limit of regenerative braking torque T_reg_max and (1-K)·T_total. This achieves a clear allocation between regenerative braking and mechanical braking torques, enabling the system to output the specific output quantities of both under given K and T_total conditions simply through formula calculation. Furthermore, the value of the regenerative braking torque always does not exceed its maximum available upper limit, while ensuring that the sum of regenerative and mechanical braking torques equals the total braking torque requirement.

[0045] Furthermore, in one specific implementation, the upper limit of the regenerative braking torque T_reg_max is determined by T_reg_max = min(T_motor_max(ω), P_chg_max(SOC, T_bat) / ω). The regenerative braking torque is determined by looking up the motor torque MAP table pre-stored in the controller's non-volatile memory, and P_chg_max is sent in real time by the battery management system (BMS) via the CAN bus. T_bat represents the battery temperature. In this specific implementation, the upper limit of regenerative power is strictly limited by the current thermal safety state of the battery. Especially when T_bat increases, causing P_chg_max to decrease, the regenerative braking torque request can be actively reduced to avoid the risk of battery overheating, lifespan degradation, or thermal runaway caused by forced high-power charging under high-temperature conditions. At the same time, the mechanical braking torque ratio coefficient K generated by slope identification and brake temperature prediction is combined to achieve coordinated allocation of regenerative and mechanical braking torques. This ensures that energy recovery efficiency is maximized and dual thermal safety of the battery and braking system is guaranteed while meeting the vehicle deceleration requirements. This solves the problem of misjudgment of charging capacity and insufficient safety redundancy caused by the upper limit of regenerative braking torque not considering the influence of battery temperature in the prior art, and realizes precise coordinated control of power output and thermal management of the regenerative braking system under complex working conditions.

[0046] Furthermore, when T_reg_max is less than (1-K)·T_total, the mechanical braking torque T_mech is compensated to T_mech = T_total - T_reg to ensure that the total braking torque accurately matches the target deceleration. When the upper limit of the regenerative braking torque T_reg_max is insufficient to meet the requirement of (1-K)·T_total, the mechanical braking torque T_mech is adjusted to T_totalT_reg through a dynamic compensation mechanism, so that the total braking torque is always precisely equal to the sum of the regenerative and mechanical braking torques. Thus, even when the regenerative braking capacity is limited, it can still fully match the total braking torque requirement determined by the slope, target deceleration, and vehicle speed, effectively avoiding deceleration deviation caused by insufficient regenerative braking capacity. Combined with the collaborative constraint mechanism of real-time slope identification and brake temperature prediction, it ensures the safety and control accuracy of the braking system under complex conditions such as high temperature and steep slope, ultimately achieving the optimal synergy between regenerative braking energy recovery efficiency and mechanical braking reliability, and improving the stability and intelligence level of the vehicle's braking performance.

[0047] In summary, the technical solution of this embodiment utilizes the vehicle's existing acceleration and wheel speed sensors to continuously estimate the road slope angle α based on the vehicle's longitudinal dynamics equations in the absence of acceleration commands. This slope angle is then used as a quantitative input in the calculation of the total braking torque T_total, ensuring that the braking force supply accurately matches the actual demand caused by the gravity component of the slope. This overcomes the mismatch between braking force supply and demand caused by relying solely on qualitative slope judgment in traditional technologies. Simultaneously, the mechanical brake temperature T_brake is predicted in real time using a lumped parameter thermal model, and the temperature is adjusted accordingly when it reaches a certain level. When the temperature reaches or exceeds 200℃, the mechanical braking torque ratio coefficient K is forcibly limited to no more than 0.3. This proactively suppresses heat generation from mechanical braking at the source of braking force distribution, preventing the risk of thermal degradation caused by continuous temperature accumulation. Furthermore, by using slope angle α, vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake as multi-dimensional input variables, a fuzzy controller dynamically outputs the K value. This achieves synergistic optimization of regenerative braking and mechanical braking under slope adaptability and thermal safety constraints, ultimately maximizing the electric braking recovery ratio while ensuring the target deceleration and guaranteeing the thermal safety of the braking system. Therefore, this effectively solves the problem of multi-factor synergistic optimization of regenerative braking and mechanical braking based on quantitative slope input and pre-constraint of brake temperature in existing technologies, achieving comprehensive technical effects such as improving energy recovery efficiency, enhancing the thermal stability of the braking system, and extending brake life.

[0048] Example 2: To achieve dynamic tracking of the temperature of a mechanical brake without the need for a dedicated temperature sensor, a computationally simple and real-time temperature prediction model is required. This application uses a discrete recursive thermal model to achieve this function.

[0049] Specifically, in step S3, the mechanical brake temperature T_brake is recursively estimated in real time. This includes: within each sampling period Ts, based on the mechanical braking torque T_mech(k-1), wheel speed ω(k-1), ambient temperature T_ambient, and thermal parameters h, A, c, and m_brake of the previous period, executing T_brake(k) = T_brake(k-1) + [T_mech(k-1)·ω(k-1) - h·A·(T_brake(k-1) - T_ambient)] / (c·m_brake) · Ts, where the sampling period Ts ≤ 200ms.

[0050] In this embodiment, within each sampling period Ts≤200ms, based on the mechanical braking torque T_mech(k-1), wheel speed ω(k-1), ambient temperature T_ambient, and thermal parameters h, A, c, and m_brake of the previous period, a discrete recursive formula is used: T_brake(k) = T_brake(k-1)[T_mech(k-1)·ω(k-1) - h·A·(T_brake(k-1) -T_ambient)] / (c·m_brake) · Ts updates the predicted temperature of the mechanical brake in real time, enabling the construction of a highly synchronized, high-frequency thermal state estimation model based on existing wheel speed and braking torque signals without adding temperature sensors. This ensures strict synchronization between temperature prediction and braking force distribution control cycle, thereby guaranteeing that when the predicted temperature reaches or exceeds 200℃, the fuzzy controller can immediately trigger a forced limiting strategy with a mechanical braking torque ratio coefficient K≤0.3. This proactively intervenes in the risk of brake overheating, avoiding thermal decay or component damage caused by temperature response lag. It effectively solves the problem that traditional systems cannot achieve pre-constraint due to temperature monitoring delays, achieving a synergistic optimization effect of accurately ensuring the thermal safety of the braking system and the collaborative efficiency of regenerative braking under low cost and without additional sensors.

[0051] In one specific embodiment of this example, the thermal parameters in the lumped parameter thermal model are: m_brake = 0.05 kg, c = 500 J / (kg·K), A = 0.002 m², h = 30 W / (m²·K), and T_ambient is collected by the vehicle ambient temperature sensor. When the vehicle is powered on and started, T_brake is initialized to the current value of T_ambient.

[0052] In this specific embodiment, the mass parameter m_brake in the lumped-parameter thermal model is limited to 0.05 kg, the specific heat capacity c is 500 J / (kg·K), the heat dissipation area A is 0.002 m², and the convective heat transfer coefficient h is 30. The system calculates the temperature using W / (m²·K) and collects the T_ambient value in real time based on the vehicle's existing ambient temperature sensor. Simultaneously, when the vehicle is powered on and started, the initial temperature T_brake of the mechanical brake is set to the current value of T_ambient. This allows the temperature prediction model to perform high-precision recursive calculations based on standardized, measured, and calibrated thermal parameters and real-world initial conditions, even without a dedicated brake temperature sensor. This effectively eliminates temperature estimation drift caused by parameter ambiguity or initial value assumption deviations. Consequently, it ensures that the safety constraint logic, which forces the mechanical braking torque ratio coefficient K to be limited to within 0.3 when T_brake ≥ 200℃, can be triggered accurately and promptly. This avoids the risk of excessive regenerative braking intervention or mechanical braking overload caused by temperature misjudgment. Ultimately, it achieves synergistic optimization of regenerative braking and mechanical braking based on slope identification and pre-constrained temperature, significantly improving the safety and reliability of the braking system in complex slope and environmental fluctuation scenarios.

[0053] Example 3: In order to ensure that each input variable can obtain a reasonable and precise fuzzy mapping during the fuzzy inference process, it is necessary to clearly define the fuzzy universe, fuzzy set partition and membership function of each input variable.

[0054] Specifically, this application sets the fuzzification parameters for each input variable of the fuzzy controller as follows:

[0055] First, the fuzzy universe of discourse for the slope angle α is [-10°, 10°], divided into five fuzzy sets: large uphill (NB, center point 8°), small uphill (NS, center point 3°), flat road (ZO, center point 0°), small downhill (PS, center point 3°), and large downhill (PB, center point 8°), with trigonometric functions as membership functions. The fuzzy controller uses the slope angle α as the key input variable and precisely limits its fuzzy universe of discourse to [-10°, 10°]. The slope angle α is divided into five fuzzy sets with clear physical meaning and center point distribution: large uphill (NB, center point 8°), small uphill (NS, center point 3°), flat road (ZO, center point 0°), small downhill (PS, center point 3°), and large downhill (PB, center point 8°). Each fuzzy set is mapped using a triangular membership function. This structure enables high-resolution, non-overlapping, and smooth transition fuzzification of the slope angle α within the typical road slope range. This significantly improves the controller's response sensitivity to small changes in slope and the output continuity of the braking torque ratio coefficient K. It avoids control abrupt changes or response lag caused by coarse input division. Combined with the collaborative constraints of battery state of charge (SOC) and brake predicted temperature (T_brake), it ensures that the K value is forcibly limited to below 0.3 under high-temperature conditions. This achieves stable, accurate, and safe coordination of regenerative braking and mechanical braking under different slope and thermal load conditions. It effectively solves the technical defects of unstable braking force distribution and discontinuous system response caused by imprecise fuzzification of slope input in traditional methods.

[0056] Second, the fuzzy domain of the vehicle speed v is [0, 30] km / h, which is divided into three fuzzy sets: low speed (L, center point 5 km / h), medium speed (M, center point 15 km / h), and high speed (H, center point 25 km / h), and the membership function is a trigonometric function. By limiting the fuzzy domain of vehicle speed v to [0, 30] km / h and dividing it into three triangular membership function fuzzy sets with center points of 5 km / h, 15 km / h, and 25 km / h respectively, the fuzzy controller forms a standardized, high-resolution fuzzification processing mechanism for vehicle speed changes within the typical driving speed range of shared electric bicycles. This ensures that, under the joint constraints of slope angle and brake temperature prediction, the mechanical braking torque ratio coefficient K can achieve precise and stable dynamic adjustment according to subtle changes in vehicle speed. This avoids control response lag or oscillation caused by a wide or unreasonable fuzzy domain. Furthermore, without weakening the temperature safety constraints and slope adaptive capability, it significantly improves the response accuracy and operational stability of the coordinated distribution of regenerative braking and mechanical braking, effectively solving the problem of uncoordinated braking force distribution caused by the lack of vehicle speed fuzzification design.

[0057] Third, the fuzzy universe of the battery's state of charge (SOC) is [0, 100]%, which is divided into three fuzzy sets: low (L, center point 20%), medium (M, center point 60%), and high (H, center point 90%), with trigonometric functions as membership functions. By explicitly defining the fuzzy domain of the battery state of charge (SOC) as [0, 100]%, and dividing it into three fuzzy sets with specific center points (low: 20%, medium: 60%, high: 90%) and using triangular membership functions, the SOC input variable has a standardized and repeatable fuzzy mapping logic in the fuzzy controller. This ensures that the numerical changes of SOC under different vehicle operating conditions can be stably and accurately converted into three fuzzy states: "low," "medium," and "high." Together with the slope angle α, vehicle speed v, and estimated temperature T_brake, it forms a four-dimensional input space to participate in fuzzy inference, realizing the dynamic and adaptive adjustment of the regenerative braking torque ratio coefficient K. In particular, when the SOC is in the low value range, mechanical braking is prioritized, and regenerative braking is prioritized in the high value range. This effectively avoids the problem of brake force distribution fluctuation and response inconsistency caused by the lack of fuzzy definition of SOC, significantly improving the accuracy, robustness, and engineering feasibility of the regenerative braking strategy under battery state change scenarios.

[0058] Fourth, the fuzzy domain of the predicted temperature T_brake is [25℃, 300℃], which is divided into three fuzzy sets: normal (N, center point 60℃), slightly high (P, center point 160℃), and dangerous (D, center point 250℃). The membership function is a trigonometric function, and when T_brake≥200℃, K≤0.3 is forcibly restricted and it does not participate in fuzzy inference. By limiting the fuzzy domain of the estimated temperature T_brake to [25℃, 300℃] and dividing it into three triangular membership function fuzzy sets with center points of 60℃, 160℃, and 250℃ respectively (normal, slightly high, and dangerous), standardized hierarchical quantization processing of the brake temperature input is achieved. This enables the fuzzy controller to respond continuously and without abrupt changes to the "normal" and "slightly high" states based on clear and smooth membership relationships when the temperature is below 200℃, effectively avoiding abnormal fluctuations in the braking torque ratio coefficient K caused by incomplete fuzzification of the temperature input. At the same time, combined with the hard constraint that K≤0.3 and does not participate in fuzzy inference when T_brake≥200℃, the control logic smoothly transitions from fuzzy inference to deterministic forced protection when the temperature approaches the critical threshold. This eliminates control discontinuities and response uncertainties caused by fuzzy inference failure or boundary fuzziness in the 200℃ to 250℃ range, thereby significantly improving the safety and control stability of the mechanical braking system under high-temperature conditions while ensuring the efficiency of regenerative braking energy recovery.

[0059] Example 4: In the fuzzy controller design in step S4, the fuzzification and defuzzification methods of the output variable K directly affect the accuracy and smoothness of the braking torque distribution. In order to obtain continuous and accurate K value output, this application sets the fuzzy set partitioning and defuzzification method of the output variable K as follows.

[0060] Specifically, the fuzzy universe of the output variable K is [0, 1], divided into five fuzzy sets: very small (VS, center point 0.05), small (S, center point 0.2), medium (M, center point 0.4), large (L, center point 0.6), and very large (VL, center point 0.85). The centroid method is used for defuzzification, and optionally, the single inference time is ≤5ms. By limiting the fuzzy universe of the output variable K to [0, 1] and dividing it into five fuzzy sets with definite center points: very small (VS, center point 0.05), small (S, center point 0.2), medium (M, center point 0.4), large (L, center point 0.6), and very large (VL, center point 0.85), combined with the centroid method defuzzification strategy, the output of K achieves a smooth transition of five-level gradients across the entire range, effectively improving the resolution and stability of braking torque allocation and avoiding sudden changes in braking force caused by coarse fuzzy set division or improper defuzzification methods. Simultaneously, through constraints... The single fuzzy inference time does not exceed 5ms, ensuring rapid dynamic adjustment within a 200ms control cycle. This meets the stringent real-time requirements of low-computing-power platforms, enabling precise, continuous, and low-latency control of the mechanical braking torque ratio under complex operating conditions involving multiple factors such as slope changes, battery state fluctuations, and brake temperature increases. This solves problems such as uncoordinated regeneration and mechanical braking, rough braking experience, and delayed thermal protection response caused by unclear output control accuracy and response speed, significantly improving the system's energy recovery efficiency and braking safety under slope conditions.

[0061] Example 5: Under conditions of continuous downhill driving or frequent braking, the brake temperature may show a continuous upward trend. In order to intervene in advance and suppress the temperature rise before the temperature reaches the 200°C forced limiting threshold, this application sets up a K-value locking mechanism for the temperature transition zone.

[0062] Specifically, in the fuzzy rule base, on the one hand, when the estimated temperature T_brake is in the range of 120℃ to 200℃, the increase of the K value is prohibited, the K value of the previous cycle is kept unchanged, and only the decrease or maintenance of the K value is allowed. It is easy to understand that when the estimated brake temperature T_brake is in the transition temperature range of 120℃ to 200℃, the control strategy of prohibiting the increase of the mechanical braking torque ratio coefficient K, forcibly maintaining the K value of the previous cycle unchanged, and only allowing the decrease or maintenance of the K value, together with the hard constraint of the mandatory limit K≤0.3 when T_brake≥200℃, forms a stepped temperature protection system. In this way, the upward trend of the mechanical braking ratio is actively suppressed before the risk of thermal fade reaches the critical point, effectively blocking the heat accumulation effect caused by the unconstrained growth of the K value, so that the brake temperature tends to stabilize in the runaway range, avoiding its continuous rise and exceeding the 200℃ threshold. This achieves forward-looking and graded prevention of the risk of thermal fade in the braking system, and significantly improves the safety margin and temperature robustness of the coordinated control of regenerative braking and mechanical braking.

[0063] On the other hand, when the estimated temperature T_brake rises from below 120°C to 120°C, the K-value locking mechanism is triggered, preventing the K-value from increasing until T_brake drops below 100°C. When the estimated brake temperature T_brake rises from below 120℃ to 120℃, the system triggers a K-value locking mechanism, preventing the mechanical braking torque ratio coefficient K from increasing further until T_brake falls below 100℃. This mechanism works in synergy with the logic of the fuzzy controller based on the slope angle α, vehicle speed v, battery state of charge (SOC), and the dynamic output K of T_brake. It blocks the unexpected rise in K value caused by small temperature fluctuations or controller response delays in the early stage of temperature rise, effectively preventing the mechanical braking ratio from being improperly amplified in the critical stage of heat accumulation. This suppresses the positive temperature feedback loop caused by the increased heat generation of the brake. Compared with the scheme that only forces K≤0.3 when T_brake is greater than or equal to 200℃, it achieves earlier intervention and more precise thermal protection control, significantly delaying the brake temperature rise trend before reaching the dangerous threshold, and improving the thermal stability and safety margin of the braking system under continuous braking conditions on long slopes.

[0064] Example 6: The regenerative braking collaborative control method based on real-time slope identification and brake temperature prediction provided in this application is particularly suitable for the specific application scenario of shared electric bicycles.

[0065] Specifically, the power supply system of the shared bicycles is a 48V lithium battery, without ABS or electronic stability system, and the control unit is an ARM Cortex-M4 core MCU with a main frequency of ≥168MHz and memory of ≥256KB. For the unique application scenario of shared electric bicycles in the absence of ABS or electronic stability systems, low-computing-power hardware platforms, and 48V lithium battery power supply architecture, this paper proposes a solution. By limiting the control unit to an ARM Cortex-M4 core MCU with a main frequency ≥168MHz and memory ≥256KB, and ensuring that all algorithm steps, including slope angle estimation, mechanical brake temperature recursive prediction, fuzzy controller input variable processing, and braking torque distribution calculation, are stably executed in real time within a 200ms cycle, this solution effectively solves the risk of control lag and inaccurate braking force distribution caused by insufficient computing power or memory constraints in traditional regenerative braking cooperative control methods on resource-constrained platforms. The deep adaptation of this hardware configuration and algorithm flow enables the cooperative control strategy based on quantitative slope input and pre-constraint of brake temperature to operate reliably in a simplified architecture without complex electronic stability systems. This ensures both braking safety and energy recovery efficiency, while avoiding system instability caused by exceeding control cycle limits, thus achieving highly robust and real-time braking cooperative control in typical shared electric bicycle application environments.

[0066] Furthermore, it is evident that the regenerative braking coordinated control method based on real-time slope identification and brake temperature prediction in this application does not require the installation of slope sensors, brake temperature sensors, or brake pressure sensors. All input parameters are obtained through software algorithms using the vehicle's existing triaxial acceleration sensors, wheel speed sensors, motor current detection circuit, and ambient temperature sensor. By utilizing the vehicle's existing triaxial acceleration sensors and wheel speed sensors, combined with real-time soft measurement of the road slope angle α using the longitudinal dynamic equation, calculating the mechanical braking power based on the motor current detection circuit, and integrating ambient temperature data obtained from the ambient temperature sensor, a lumped parameter thermal model is constructed to recursively estimate the mechanical brake temperature T_brake. Simultaneously, the vehicle speed v is directly obtained from the wheel speed sensor. Thus, without the need to add slope sensors, brake temperature sensors, or brake pressure sensors, all parameters such as slope angle α, vehicle speed v, battery state of charge (SOC), and predicted temperature T_brake are completely obtained. By controlling the input parameters, the fuzzy controller can dynamically output the mechanical braking torque ratio coefficient K based on these soft measurement parameters, and automatically force the amplitude to be limited to K≤0.3 when T_brake≥200℃. This allows for precise allocation of regenerative braking torque and mechanical braking torque, achieving coordinated optimization control of regenerative braking and mechanical braking without dedicated sensors. This effectively overcomes the practical bottleneck of difficult deployment of control strategies in cost-sensitive and space-constrained platforms such as shared electric bicycles due to difficulties in hardware upgrades. It achieves the goal of high-precision and high-safety braking energy recovery and thermal protection coordination even with low hardware configuration.

[0067] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0068] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, or improvements made within the principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A regenerative braking coordinated control method based on real-time slope identification and brake temperature prediction, characterized in that, include: S1. In the absence of acceleration command, based on the acquired longitudinal acceleration a_x and vehicle speed v, the road slope angle α is obtained through iterative calculation using the vehicle longitudinal dynamics equation; The expression for the longitudinal dynamic equation of the vehicle is: sinα = (m·a_x - μ·m·g·cosα - 0.5·ρ·Cd·A·v²) / (m·g), where m is the vehicle mass, μ is the rolling resistance coefficient, ρ is the air density, Cd is the drag coefficient, A is the vehicle's frontal area, and the vehicle mass m is obtained by inverse calculation of air resistance and rolling resistance under no acceleration command conditions; S2. Based on the slope angle α, the preset target deceleration a_target and the vehicle speed v, calculate the total braking torque T_total required to maintain the preset target deceleration using the vehicle dynamics formula; The formula for calculating the total braking torque T_total is: T_total = [m·|a_target| + m·g·sinα +μ·m·g·cosα + 0.5·ρ·Cd·A·v²] · r, where the preset target deceleration a_target ranges from 0.8 m / s² to 2.0 m / s², r is the wheel rolling radius, and when α≥0, m·g·sinα is added as a positive braking force component, and when α≤0, m·g·sinα is subtracted as a negative braking force component; S3. Based on the lumped parameter thermal model of mechanical braking power, heat dissipation parameters and ambient temperature, the mechanical brake temperature T_brake is predicted in real time using recursive calculation. S4. Using the slope angle α, vehicle speed v, battery state of charge (SOC) and estimated temperature T_brake as input variables, the mechanical braking torque ratio coefficient K is output through a fuzzy controller. When the estimated temperature T_brake ≥ 200℃, K is forcibly limited to ≤ 0.

3. The input variables of the fuzzy controller include slope angle α, vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake. The core rule of the fuzzy rule base corresponding to the fuzzy controller is as follows: when the slope angle α is a steep downhill slope (PB), the vehicle speed v is high (H), the battery state of charge (SOC) is low (L), and the estimated temperature T_brake is normal (N), the mechanical braking torque ratio coefficient K is output as very small (VS); when the estimated temperature T_brake is high (P), regardless of the values ​​of slope angle α, vehicle speed v, and battery state of charge (SOC), the mechanical braking torque ratio coefficient K is output as no more than 0.3; when the slope angle α is a steep uphill slope (NB), regardless of the values ​​of vehicle speed v, battery state of charge (SOC), and estimated temperature T_brake, the mechanical braking torque ratio coefficient K is output as very large (VL). S5. Calculate the regenerative braking torque T_reg = min(T_reg_max, (1-K)·T_total) based on the mechanical braking torque ratio coefficient K and the total braking torque T_total, and calculate the mechanical braking torque T_mech = K·T_total; The upper limit of the regenerative braking torque T_reg_max is determined by T_reg_max = min(T_motor_max(ω), P_chg_max(SOC, T_bat) / ω), where T_motor_max(ω) is obtained by looking up the motor generating torque MAP table pre-stored in the controller's non-volatile memory, P_chg_max is sent in real time by the battery management system (BMS) via the CAN bus, and T_bat is the battery temperature; when T_reg_max is less than (1-K)·T_total, the mechanical braking torque T_mech is compensated to T_mech = T_total - T_reg to ensure that the total braking torque accurately matches the target deceleration.

2. The regenerative braking coordinated control method according to claim 1, characterized in that, The real-time recursive estimation of the mechanical brake temperature T_brake specifically includes: Within each sampling period Ts, based on the mechanical braking torque T_mech(k-1), wheel speed ω(k-1), ambient temperature T_ambient, and thermal parameters h, A, c, and m_brake from the previous period, the following is executed: T_brake(k) = T_brake(k-1) + [T_mech(k-1)·ω(k-1) - h·A·(T_brake(k-1) - T_ambient)] / (c·m_brake) · Ts, where the sampling period Ts ≤ 200ms; The thermal parameters in the lumped parameter thermal model are: m_brake = 0.05 kg, c = 500 J / (kg·K), A = 0.002 m², h = 30 W / (m²·K). T_ambient is collected by the vehicle ambient temperature sensor, and T_brake is initialized to the current value of T_ambient when the vehicle is powered on and started.

3. The regenerative braking coordinated control method according to claim 1, characterized in that, The fuzzy domain of the slope angle α is [-10°, 10°], which is divided into five fuzzy sets: large uphill (NB, center point 8°), small uphill (NS, center point 3°), flat road (ZO, center point 0°), small downhill (PS, center point 3°), and large downhill (PB, center point 8°), and the membership function is a trigonometric function; The fuzzy domain of the vehicle speed v is [0, 30] km / h, which is divided into three fuzzy sets: low speed (L, center point 5 km / h), medium speed (M, center point 15 km / h), and high speed (H, center point 25 km / h), and the membership function is a trigonometric function; The fuzzy universe of discourse for the battery's state of charge (SOC) is [0, 100]%, divided into three fuzzy sets: low (L, center point 20%), medium (M, center point 60%), and high (H, center point 90%), with trigonometric functions as membership functions; The fuzzy universe of the estimated temperature T_brake is [25℃, 300℃], which is divided into three fuzzy sets: normal (N, center point 60℃), slightly high (P, center point 160℃), and dangerous (D, center point 250℃). The membership function is a triangular function, and when T_brake≥200℃, K≤0.3 is forcibly restricted and it does not participate in fuzzy inference.

4. The regenerative braking coordinated control method according to claim 1, characterized in that, The fuzzy universe of the output variable K is [0, 1], which is divided into five fuzzy sets: very small (VS, center point 0.05), small (S, center point 0.2), medium (M, center point 0.4), large (L, center point 0.6), and very large (VL, center point 0.85). The centroid method is used for defuzzification.

5. The regenerative braking coordinated control method according to claim 1, characterized in that, In the fuzzy rule base, when the estimated temperature T_brake is in the range of 120℃ to 200℃, the K value is prohibited from increasing and the K value of the previous cycle is kept unchanged. Only the K value is allowed to decrease or remain unchanged. When the estimated temperature T_brake rises from below 120℃ to 120℃, the K-value locking mechanism is triggered, preventing the K-value from increasing until T_brake drops below 100℃ before the lock is released.

6. The regenerative braking coordinated control method according to claim 1, characterized in that, The method is applicable to shared electric bicycles, whose drive system is a hub-type brushless DC motor.

Images