A method for vehicle stability control in the event of failure of an EMB braking system

Abstract

The application provides a kind of EMB brake system failure vehicle stability control method, it is related to vehicle brake system technical field, specific steps include: obtaining historical data to construct training sample;Through state discrimination model, determine EMB system is normal or failure state;When failure, identify driver " slight / normal / emergency " brake intention, match corresponding wheel braking torque initial constraint interval;Combined with steering wheel angle, divide driving state, dynamically adjust constraint interval, generate initial population and predict lateral acceleration and braking distance;Build comprehensive evaluation coefficient, minimize as target optimization to obtain optimal torque combination;According to optimal combination, distribute brake force.The application is through " intention grading adaptation + state dynamic correlation " strategy, makes brake force distribution accurate response driver demand and adapts vehicle safety boundary, improves the control effect and working condition matching degree when EMB failure.

Description

Technical Field

[0001] This invention relates to the field of vehicle braking system technology, specifically to a method for controlling vehicle stability in the event of EMB braking system failure. Background Technology

[0002] Electromechanical braking (EMB) systems are a new generation of braking technology that replaces traditional hydraulic braking. They generate braking torque directly by driving the brake calipers with an electric motor, eliminating the need for hydraulic lines. Compared to traditional hydraulic braking, EMB systems offer advantages such as faster response, more compact structure, higher control precision, and easier integration with autonomous driving systems. They are considered one of the core braking technologies for intelligent connected vehicles and new energy vehicles, and are widely used in vehicle scenarios requiring high braking response speed and control flexibility. When an EMB system experiences brake failure on one or more wheels, it can lead to an imbalance in the distribution of braking torque among the wheels—for example, a sudden drop in braking torque after the motor on one wheel fails can cause problems such as vehicle skidding, yaw, and increased braking distance, seriously threatening driving safety.

[0003] The prior art, disclosed in CN120645918A, provides a method, system, device, and medium for assisting in the failure handling of a vehicle braking system. The method includes the following steps: determining the failure state of the electro-hydraulic braking system; if the system is not in a failure state, executing a pressure-building braking strategy; if the system is in a failure state, executing an auxiliary processing strategy. The auxiliary processing strategy includes: determining the braking intention based on vehicle braking feedback; based on the determination result, sequentially setting the vehicle stability control threshold, distributing the deceleration between electric and hydraulic braking, actively adjusting the braking pressure, and actively adjusting the motor torque; and monitoring the vehicle's wheel slip ratio in real time, and correcting the vehicle's steering force in conjunction with the vehicle braking feedback. This method can ensure that the vehicle deceleration meets the driver's needs while preventing the vehicle from fishtailing or overturning due to sudden steering wheel movements, thus ensuring vehicle stability.

[0004] However, the following shortcomings still exist. As can be seen from the above statement, the existing technology adopts a "fixed parameter strategy" for braking force distribution in the failure state. The uniform braking torque range and distribution ratio do not take into account the driver's differentiated braking intentions from slight deceleration to emergency avoidance. The pre-set braking torque constraint range is only based on static parameters such as the maximum load capacity of the wheel or the average road surface adhesion coefficient, and is not related to the real-time operating state such as vehicle steering and attitude. This combination of "lack of intention adaptation" and "insufficient state correlation" makes it impossible for the braking force distribution to meet the driver's needs for braking performance or comfort in different scenarios, and it is also difficult to dynamically adapt to the vehicle's real-time changing safety boundaries, ultimately leading to a disconnect between the control effect and the actual working conditions.

[0005] The information disclosed in the background section is only intended to enhance the understanding of the background of this disclosure, and therefore may include information that does not constitute prior art known to those skilled in the art. Summary of the Invention

[0006] The purpose of this invention is to provide a vehicle stability control method in the event of EMB braking system failure, so as to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution:

[0008] A method for vehicle stability control in the event of EMB braking system failure, comprising the following steps:

[0009] S1. Obtain historical data of braking torque of each wheel and vehicle dynamic parameters under normal and failure states, extract the temporal features of the historical data, construct a multi-dimensional feature vector, and form training samples.

[0010] S2. Using a state discrimination model trained based on training samples, the multidimensional feature vector of the current time period is discriminated, and the corresponding working state of the EMB braking system is output, which is either normal or failed.

[0011] S3. When the condition is determined to be normal, the braking torque of each wheel is allocated according to the original allocation rules; when the condition is determined to be in failure, the current vehicle pedal parameters are collected, the corresponding driver braking intention category is identified, and the corresponding initial constraint range of braking torque for each wheel is matched.

[0012] S4. Based on the vehicle steering wheel angle, dynamically adjust the upper and lower limits of the initial constraint range of braking torque of each wheel, randomly generate multiple sets of wheel braking torque combinations within the adjusted constraint range to construct an initial population, and use the index prediction model trained based on wheel braking torque to predict the lateral acceleration and braking distance corresponding to each wheel braking torque combination.

[0013] S5. Process lateral acceleration and braking distance to generate comprehensive evaluation coefficients. With minimizing the comprehensive evaluation coefficients as the objective function, use an optimization algorithm to iteratively optimize the wheel braking torque combination to obtain the optimal wheel braking torque combination.

[0014] S6. Distribute the braking torque of each wheel according to the optimal wheel braking torque combination.

[0015] Furthermore, the vehicle dynamic parameters include vehicle speed and vehicle yaw rate; the pedal parameters include brake pedal travel, brake pedal force, and brake pedal speed.

[0016] Furthermore, the temporal features of historical data are extracted to construct a multi-dimensional feature vector. The specific logic is as follows:

[0017] The collected historical data includes the EMB braking system operating status at each historical moment;

[0018] For each historical moment, a time period is pushed back, and within that time period, the temporal characteristics of the braking torque of each wheel and the vehicle dynamic parameters are collected to form a multi-dimensional feature vector.

[0019] Establish a mapping relationship between the multidimensional feature vector of each time period and the working state of the EMB braking system at the corresponding historical moment.

[0020] Furthermore, the braking torque of each wheel is allocated according to the original allocation rules, with the specific logic as follows:

[0021] Calculate the total braking force required based on the driver's braking request and the vehicle's current speed;

[0022] Obtain the vertical load of each wheel and the coefficient of friction between the tire and the road surface, and calculate the maximum braking force that each wheel can withstand.

[0023] Based on the preset front and rear axle and left and right wheel distribution ratios, and combined with the maximum braking force that each wheel can bear, the total braking force is distributed, and the corresponding braking torque is delivered to the braking actuators of each wheel based on the tire radius.

[0024] Furthermore, the driver's braking intention categories include light braking, normal braking, and emergency braking;

[0025] Define the first preset threshold values ​​for brake pedal travel, brake pedal force, and brake pedal speed as follows: , , The second preset threshold values ​​for brake pedal travel, brake pedal force, and brake pedal speed are respectively... , , ;

[0026] If any of the above pedal parameters of the current vehicle exceeds the corresponding second preset threshold, then emergency braking is initiated.

[0027] If none of the above pedal parameters of the current vehicle exceed the corresponding second preset threshold, and any pedal parameter is between the first preset threshold and the second preset threshold, then it is normal braking.

[0028] If none of the pedal parameters of the current vehicle exceed the corresponding first preset threshold, then it is considered light braking.

[0029] Furthermore, based on the current type of driver braking intention, the corresponding initial constraint range for braking torque of each wheel is matched, with the specific logic as follows:

[0030] Define the initial constraint range of braking torque for each wheel. ;

[0031] in, The first Initial constraint range of braking torque for each wheel. The first The lower and upper limits of the initial constraint range for the braking torque of each wheel, where:

[0032] ;

[0033] ;

[0034] in, The figures are respectively under light braking, normal braking, and emergency braking. The lower limit of the initial constraint range for the braking torque of each wheel. The figures are respectively under light braking, normal braking, and emergency braking. The upper limit of the initial constraint range for the braking torque of each wheel;

[0035] For any wheel, the following relationship is satisfied:

[0036] ;

[0037] ;

[0038] in, For the index of the wheel, , These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0039] Based on the current driver braking intention category, obtain the corresponding initial constraint range of braking torque for each wheel.

[0040] Furthermore, the upper and lower limits of the braking torque constraint range for each wheel are dynamically adjusted based on the vehicle's steering wheel angle. The specific logic is as follows:

[0041] When the steering wheel angle is less than the first preset angle threshold, it is defined as straight driving or slight steering.

[0042] When the steering wheel angle is greater than or equal to the first preset angle threshold and less than or equal to the second preset angle threshold, it is defined as moderate steering.

[0043] When the steering wheel angle is greater than the second preset angle threshold, it is defined as a sharp turn;

[0044] Specifically, the upper and lower limits of the initial constraint range for the braking torque of each wheel are adjusted for different steering states, as follows:

[0045] When driving in a straight line or making a slight turn, the initial constraint range of the braking torque of each wheel remains unchanged;

[0046] When steering in a moderate position, set the first steering wheel angle. :

[0047] ;

[0048] in, The first preset angle threshold, The second preset angle threshold;

[0049] The formula used to calculate the amplitude adjustment factor under moderate steering conditions is as follows:

[0050] ;

[0051] in, This is the amplitude adjustment coefficient under moderate steering conditions. The current steering wheel angle;

[0052] The formula used to calculate the upper limit of the initial constraint range for the adjusted braking torque of each wheel is as follows:

[0053] ;

[0054] in, For the adjusted number The upper limit of the initial constraint range for the braking torque of each wheel. The first preset for moderate steering conditions The adjustment range of the upper limit value for each wheel;

[0055] When making a sharp turn, set the second steering wheel angle. :

[0056] ;

[0057] The formula used to calculate the amplitude adjustment coefficient during a sharp turn is as follows:

[0058] ;

[0059] in, This is the amplitude adjustment coefficient during sharp turns. This is the maximum permissible value for the steering wheel angle;

[0060] During a sharp turn, the upper limit of the initial constraint range of the wheel braking torque, adjusted under moderate steering conditions, is further lowered to obtain the upper limit of the initial constraint range of the wheel braking torque under sharp turning conditions. The formula used is as follows:

[0061] ;

[0062] in, For the adjusted number The upper limit of the initial constraint range for the braking torque of each wheel. The preset first step for sharp turns The adjustment range of the upper limit value for each wheel;

[0063] Based on the current steering state, obtain the upper and lower limits of the corresponding adjusted braking torque constraint range for each wheel.

[0064] Furthermore, multiple combinations of wheel braking torques are randomly generated within the adjusted constraint interval to construct the initial population. The specific logic is as follows:

[0065] Within the constrained range of the adjusted braking torque of each wheel, the braking torques of the left front wheel, right front wheel, left rear wheel, and right rear wheel are randomly generated to form an initial population. This initial population is then labeled as... , , The first in the initial population Individual, This is the index of an individual in the initial population, and , The initial population size. ,in, The first The braking torque of the left front wheel, right front wheel, left rear wheel, and right rear wheel of each individual.

[0066] Furthermore, the lateral acceleration and braking distance are normalized, and a comprehensive evaluation coefficient is generated using the normalized lateral acceleration and braking distance. The formula used is as follows:

[0067] ;

[0068] ;

[0069] in, For the first The comprehensive evaluation coefficient of an individual For the first The lateral acceleration deviation value of each individual. For the first Braking distance of each individual;

[0070] In the formula, For the first The lateral acceleration of each individual, This represents the lower limit of the suitable range for lateral acceleration. This represents the upper limit of the suitable range for lateral acceleration.

[0071] In the formula, , These are the weighting coefficients for the lateral acceleration deviation and the braking distance, respectively. , The specific value is determined by the analytic hierarchy process (AHP).

[0072] Furthermore, with the minimization of the comprehensive evaluation coefficient as the objective function, an optimization algorithm is used to iteratively optimize the wheel braking torque combination to obtain the optimal wheel braking torque combination. The specific logic is as follows:

[0073] Within the constraint range of braking torques for the left front wheel, right front wheel, left rear wheel, and right rear wheel, iterative optimization is performed on individuals in the initial population. Specifically, this involves adjusting the comprehensive evaluation coefficient... Sort by size from largest to smallest, and select the comprehensive evaluation coefficient. Individuals at the top of the hierarchy serve as parents. Through crossover and mutation operations in a genetic algorithm, the gene codes of these parents are exchanged, combined, and mutated to generate new individuals. An index prediction model is used to predict the lateral acceleration and braking distance of these new individuals, and the corresponding comprehensive evaluation coefficient is calculated. The newly generated individuals are then merged with their parents to form the next generation population. The selection, crossover, and mutation operations are repeated. When the convergence condition is met, the individual with the minimum comprehensive evaluation coefficient is selected as the optimal wheel braking torque combination, denoted as... ,in, These are the braking torques of the left front wheel, right front wheel, left rear wheel, and right rear wheel in the optimal wheel braking torque combination.

[0074] Compared with the prior art, the beneficial effects of the present invention are:

[0075] This invention accurately identifies the driver's braking intention category by pre-setting three types of driver braking intentions: "slight braking," "normal braking," and "emergency braking," and combining them with the current pedal parameters. It then matches each wheel with a unique initial constraint range for braking torque, achieving precise adaptation of braking force distribution to driver needs. This meets the diverse braking scenario requirements, from "smooth deceleration" in congested traffic to "emergency avoidance" in sudden danger, and solves the problem that existing technologies cannot dynamically adjust braking force according to driver intentions.

[0076] By utilizing the steering wheel angle, the vehicle's driving state is divided into three categories: "straight line / slight steering, moderate steering, and sharp turn". The initial constraint range of braking torque is further dynamically adjusted according to the steering risk level, so that the braking torque is precisely matched with the wheel adhesion limit under the current steering. This solves the problem of the disconnect between the braking force distribution of existing technologies and the real-time operating state of the vehicle, and improves the vehicle stability in failure scenarios from the bottom up.

[0077] The adjusted braking torque constraint range provides a precise "safe and effective boundary" for generating multiple sets of wheel braking torque combinations in the future.

[0078] In summary, this invention breaks through the control bottleneck caused by the superposition of "lack of intent and disconnection of state" in the prior art through a two-layer optimization strategy of "intent hierarchical adaptation + dynamic state association". This allows the braking force distribution to not only accurately respond to the driver's braking needs, but also dynamically fit the vehicle's real-time safety boundary, ultimately significantly improving the control effect and matching degree of actual working conditions when the EMB braking system fails. Attached Figure Description

[0079] Figure 1 This is a schematic diagram of the overall method flow of the present invention;

[0080] Figure 2 This is a schematic diagram showing the fitting of the comprehensive evaluation coefficients and the lateral acceleration deviation values ​​of this invention;

[0081] Figure 3 This is a schematic diagram showing the fitting of the comprehensive evaluation coefficient and braking distance in this invention. Detailed Implementation

[0082] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to specific embodiments.

[0083] It should be noted that, unless otherwise defined, the technical or scientific terms used in this invention should have the ordinary meaning understood by one of ordinary skill in the art to which this invention pertains. The terms "first," "second," and similar terms used in this invention do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed following the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are used only to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0084] Example:

[0085] Please see Figures 1-3 The present invention provides a technical solution:

[0086] A method for vehicle stability control in the event of EMB braking system failure, comprising the following steps:

[0087] S1. Obtain historical data of braking torque of each wheel and vehicle dynamic parameters under normal and failure states, extract the temporal features of the historical data, construct a multi-dimensional feature vector, and form training samples.

[0088] Based on the above embodiments, the vehicle dynamic parameters include vehicle speed and vehicle yaw rate.

[0089] The methods for collecting the braking torque of the left front wheel, the braking torque of the right front wheel, the braking torque of the left rear wheel, the braking torque of the right rear wheel, vehicle speed, and vehicle yaw rate are as follows:

[0090] Torque sensors or torque sensors are installed on the braking actuators of each wheel to measure the braking torque of the left front wheel, right front wheel, left rear wheel, and right rear wheel in real time; vehicle speed is collected by a vehicle speed sensor; and vehicle yaw rate is measured by a gyroscope sensor in the onboard inertial measurement unit.

[0091] To further clarify, the above data needs to be normalized, and all subsequent processing and analysis of the above data will be based on the normalized data.

[0092] Based on the above embodiments, time-series features of historical data are extracted to construct a multi-dimensional feature vector. The specific logic is as follows:

[0093] The collected historical data includes the EMB braking system operating status at each historical moment;

[0094] For each historical moment, a time period is pushed back. Within this time period, the average values ​​of braking torque of each wheel, vehicle speed, and vehicle yaw rate are collected to construct a multidimensional feature vector for the historical time period. :

[0095] ;

[0096] in, This represents the average braking torque of the left front wheel over a historical period. This represents the average braking torque of the right front wheel over a historical period. This represents the average braking torque of the left rear wheel over a historical period. This represents the average braking torque of the right rear wheel over a historical period. This represents the average vehicle speed over a historical period. The average yaw rate of the vehicle over a historical period;

[0097] Establish a mapping relationship between the multidimensional feature vector of each historical time period and the working state of the EMB braking system at the corresponding historical moment.

[0098] S2. Using a state discrimination model trained based on training samples, the multidimensional feature vector of the current time period is discriminated, and the corresponding working state of the EMB braking system is output, which is either normal or failed.

[0099] Based on the above embodiments, the state discrimination model is constructed using a deep learning network based on a multilayer perceptron. The deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, a third hidden layer, and an output layer. The first hidden layer, the second hidden layer, and the third hidden layer each have at least two neurons and all use ReLU (linear rectified unit) as the activation function.

[0100] The multidimensional feature vectors from multiple historical time periods and the working state of the EMB braking system are divided into training, validation, and test sets, with a common ratio of 7:2:1. The training set is used to learn the model parameters; the validation set is used to adjust hyperparameters during training to prevent overfitting; and the test set is used to evaluate the generalization ability of the model after training.

[0101] The structure of a deep learning network with a multilayer perceptron includes:

[0102] Input layer: used to receive multidimensional feature vectors;

[0103] The first hidden layer has 128 neurons and uses ReLU as the activation function.

[0104] The second hidden layer has 64 neurons and also uses the ReLU activation function;

[0105] The third hidden layer has 32 neurons and uses the ReLU activation function;

[0106] Output layer: has 1 neuron, used to output the working status of the EMB braking system.

[0107] The output layer uses the Sigmoid function for binary classification; the three-layer structure captures the deep mapping relationship between the working state of the EMB system and multi-dimensional features more accurately than the 1-2 hidden layers through feature abstraction from "simple to complex".

[0108] The process of training the state discrimination model is as follows:

[0109] Using multidimensional feature vectors from multiple historical time periods as input and the corresponding EMB braking system operating status as the output label, a feature association model is trained. The cross-entropy loss function is used, and the training set loss and validation set loss are both ≤ Stop training the feature association model, where, This is the maximum allowable threshold for the loss difference. A value of 0.01-0.1 is used, which eliminates both overfitting and underfitting.

[0110] Based on the above embodiments, a multidimensional feature vector for the current time period is constructed. :

[0111] ;

[0112] in, This represents the average braking torque of the left front wheel during the current time period. This represents the average braking torque of the right front wheel during the current time period. This represents the average braking torque of the left rear wheel during the current time period. This represents the average braking torque of the right rear wheel during the current time period. This represents the average vehicle speed for the current time period. This represents the average yaw rate of the vehicle during the current time period.

[0113] Based on the above embodiments, the multi-dimensional feature vector of the current time period is input into the trained state discrimination model, and the corresponding EMB braking system working state is output, which is either normal or failed.

[0114] S3. When the condition is determined to be normal, the braking torque of each wheel is allocated according to the original allocation rules; when the condition is determined to be in failure, the current vehicle pedal parameters are collected, the corresponding driver braking intention category is identified, and the corresponding initial constraint range of braking torque for each wheel is matched.

[0115] Based on the above embodiments, the braking torque of each wheel is allocated according to the original allocation rules, and the specific logic is as follows:

[0116] Based on the driver's braking request and the vehicle's current speed, the required total braking force is calculated using the following formula:

[0117] ;

[0118] ;

[0119] in, For total braking force, For the total mass of the vehicle. It is the acceleration due to gravity. This represents the percentage of pedal travel. A function of the speed correction factor. For adjustment coefficients, , For vehicle speed;

[0120] When a vehicle leaves the factory, the manufacturer has already entered the "curb weight in empty condition" into the vehicle controller's database as an initial baseline value. Using seat pressure sensors, it detects whether the seats are occupied and the number of occupants, estimating the additional load based on the standard weight per person, and thus the total vehicle weight. It equals the sum of the basic curb weight and the additional load;

[0121] The pedal travel percentage is the ratio of the "actual travel of the pedal being pressed" to the "maximum effective travel of the pedal". Through a non-contact Hall sensor, the sensor is linked with the pedal linkage. When the pedal is pressed, the linkage drives the magnetic core or sliding contact inside the sensor to move, and outputs an electrical signal that is proportional to the travel. The "maximum effective travel of the pedal" is preset. The sensor collects the actual travel of the pedal at the moment and uses the ratio of it to the maximum effective travel of the pedal as the pedal travel percentage.

[0122] Electromagnetic or Hall effect wheel speed sensors are installed at the hubs of all four wheels to sense the real-time rotational angular velocity of the wheels. For each revolution of the wheel, the sensor outputs a fixed number of pulse signals. The controller calculates the wheel angular velocity by counting the number of pulses per unit time and multiplies it by the tire radius as the wheel speed.

[0123] It is an "exponential growth curve", whose growth rate is determined by... Decision, when When it increases, It gradually approaches 1, which meets the requirement of "the braking force correction range tends to stabilize at high speeds";

[0124] when When decreasing, Rapid reduction aligns with the requirement of "reduced braking force correction at low speeds to prioritize braking response";

[0125] and The range of values This allows the "growth rate" of the curve to match the actual vehicle speed range.

[0126] For the For each wheel, calculate the maximum braking force it can withstand based on the vertical load and the tire-road friction coefficient:

[0127] ;

[0128] in, For the first The maximum braking force that each wheel can withstand. For the first The coefficient of friction between the tires and the road surface of each wheel. For the first Vertical load on each wheel;

[0129] Among them, the wheel speed sensor of each wheel continuously measures the wheel speed and calculates the linear velocity of the wheel; the brake pressure sensor measures the pressure of the brake cylinder of each wheel and calculates the actual applied braking torque; the controller compares the "wheel speed" and "actual vehicle speed" to determine whether the wheel is slipping; and combines the "braking torque" and "degree of slippage" to calculate the friction coefficient between the tire and the road surface using the preset dynamic formula "friction force equals braking torque / slippage correction coefficient".

[0130] When the vehicle leaves the factory, the static base value of each wheel in the empty state is measured by special equipment and stored in the controller. The acceleration sensor installed on the vehicle body measures the longitudinal acceleration and lateral acceleration of the vehicle. The controller calculates the dynamic transfer amount based on the longitudinal acceleration and lateral acceleration. The real-time vertical load of each wheel is equal to the sum of the static base value and the dynamic transfer amount.

[0131] Based on the preset front-to-rear axle and left-to-right wheel distribution ratios, and combined with the maximum braking force that each wheel can withstand, the total braking force is distributed. The specific logic is as follows:

[0132] Based on the vehicle's design, the front axle and left / right wheel braking force distribution ratios are set as follows: The total braking force ratio of the rear axle is ,satisfy:

[0133] ;

[0134] The left and right wheel distribution ratios are respectively and ,satisfy:

[0135] ;

[0136] Calculate the braking force distributed to the front axle Braking force distributed to the rear axle :

[0137] ;

[0138] ;

[0139] Calculate the total braking force distributed between the left and right wheels:

[0140] ;

[0141] ;

[0142] ;

[0143] ;

[0144] in, , , , The total braking force is distributed to the left front wheel, right front wheel, left rear wheel, and right rear wheel respectively;

[0145] To prevent exceeding tire limits, the distributed braking force must be limited:

[0146] ;

[0147] ;

[0148] in, For the first Preliminary calculated braking force for each wheel. For the first The allocation ratio of each wheel to its axle, if For the front axle wheels, If it is a rear axle wheel, , For the first The allocation ratio of each wheel to its side, if For the left wheel, If it is the right wheel, ;

[0149] in, For the first The final braking force distributed to each wheel;

[0150] Based on the tire radius, the braking force is converted into braking torque:

[0151] ;

[0152] in, For the first Braking torque of each wheel This is the tire radius.

[0153] Based on the above embodiments, the driver's braking intention categories include light braking, normal braking, and emergency braking;

[0154] Define the first preset threshold values ​​for brake pedal travel, brake pedal force, and brake pedal speed as follows: , , The second preset threshold values ​​for brake pedal travel, brake pedal force, and brake pedal speed are respectively... , , ;

[0155] If any of the above pedal parameters of the current vehicle exceeds the corresponding second preset threshold, then emergency braking is initiated.

[0156] If none of the above pedal parameters of the current vehicle exceed the corresponding second preset threshold, and any pedal parameter is between the first preset threshold and the second preset threshold, then it is normal braking.

[0157] If none of the pedal parameters of the current vehicle exceed the corresponding first preset threshold, then it is considered light braking.

[0158] Specifically, the brake pedal travel must be greater than the pedal's mechanical free travel to ensure... The corresponding travel distance is sufficient to trigger the initial response of the braking system;

[0159] The brake pedal force must be greater than the brake pedal return spring force and the system frictional resistance to ensure... The corresponding pedal force can overcome mechanical resistance, causing the braking system to begin to generate a slight braking torque;

[0160] The speed of the brake pedal must be higher than the speed of disturbances such as vibrations during vehicle operation and slight movements of the driver's foot, so as to avoid misinterpreting slight foot movements that are not active braking as slight braking.

[0161] Method for determining the first preset threshold of each pedal parameter: Statistically analyze the pedal operation data of most drivers in scenarios such as congested following and slow deceleration, and take the 10%-20% percentile of pedal travel, pedal force and pedal speed in these scenarios as the first preset threshold.

[0162] Normal braking is for routine deceleration (such as deceleration at intersections or returning to position after overtaking), and it needs to be clearly distinguished from the functional scenarios of light braking and emergency braking. The threshold should correspond to the upper limit of "normal braking intensity".

[0163] Working backward from braking performance: Through real-vehicle testing, we determined the typical range of daily comfortable deceleration when the vehicle produces a deceleration of 0.3-0.8g, which is neither too slow nor causes passengers to lean forward. The corresponding range of pedal travel, pedal force, and pedal speed was determined. The upper limit of this range was taken as the second preset threshold to ensure that the effect of normal braking does not exceed the intensity of emergency braking.

[0164] Method for determining the second preset threshold for each pedal parameter: Statistically analyze the pedal operation data of most drivers during normal deceleration, and take the 70%-80% percentile of pedal travel, pedal force, and pedal speed in this scenario as the second preset threshold.

[0165] Based on the above, further explanation is needed:

[0166] A driver's braking action is essentially a direct response to the "desire to decelerate." Among the three parameters—brake pedal travel, pedal force, and pedal speed—the maximum parameter value is the most intuitive indicator of the "intensity of the demand." Specifically, even if other parameters are in the lower range, as long as a certain parameter exceeds the corresponding first or second preset threshold, the driver's true braking intention is determined based on the preset threshold corresponding to the highest parameter value.

[0167] This judgment logic, centered on the "highest parameter value," provides clear guidance for the braking system's response strategy: different braking intentions correspond to different system responses (for example, light braking requires only a slight braking force, while emergency braking requires triggering safety functions such as ABS). By prioritizing matching the intention category with the "highest parameter value," it ensures that the system response intensity accurately covers the driver's "highest needs"—avoiding insufficient response due to overly conservative judgments, and especially ensuring safety in emergency braking scenarios.

[0168] Based on the above embodiments, the initial constraint range of braking torque for each wheel is matched according to the current driver's braking intention category. The specific logic is as follows:

[0169] Define the initial constraint range of braking torque for each wheel. ;

[0170] in, The first Initial constraint range of braking torque for each wheel. The first The lower and upper limits of the initial constraint range for the braking torque of each wheel, where:

[0171] ;

[0172] ;

[0173] in, The figures are respectively under light braking, normal braking, and emergency braking. The lower limit of the initial constraint range for the braking torque of each wheel. The figures are respectively under light braking, normal braking, and emergency braking. The upper limit of the initial constraint range for the braking torque of each wheel;

[0174] Further explanation:

[0175] Through real-vehicle testing, it was determined that when the vehicle experiences a deceleration of 0.1g-0.3g, the first... The braking torque range corresponding to each wheel. The upper and lower limits of this range are taken as... and This ensures that the torque output can only trigger "very gentle deceleration" without affecting driving comfort.

[0176] in, It must be higher than the "minimum effective torque of the braking system" to ensure that even slight braking produces a perceptible deceleration effect and avoids "operational ineffectiveness".

[0177] Through real-vehicle testing, it was determined that the vehicle produces a deceleration of 0.3g-0.8g, which is the mainstream range for comfortable deceleration in daily life. Passengers experience a slight sensation but no forward tilting. The braking torque range corresponding to each wheel is taken as the upper and lower limits of the range. and This ensures that the torque output is adapted to "normal deceleration," neither too slow nor too abrupt.

[0178] The maximum coefficient of adhesion between the tire and the road surface is the core factor limiting deceleration.

[0179] The peak adhesion coefficient of dry asphalt pavement is approximately 0.8-1.0, corresponding to a maximum deceleration of approximately 0.8g-1.0g.

[0180] On a few high-adhesion surfaces, the grip weight can reach 1.0g-1.2g, but this is close to the tire's grip limit.

[0181] If the deceleration exceeds 1.2g, even under optimal road conditions, the tires will slip due to insufficient traction, leading to frequent ABS intervention or even ABS failure. This, in turn, increases braking distance and increases the risk of skidding. Therefore, 1.2g is the safety threshold in most scenarios.

[0182] The vehicle is determined to experience a deceleration of 0.8g-1.2g, which is the emergency deceleration range. Therefore, it is necessary to quickly shorten the braking distance. The braking torque range corresponding to each wheel is taken as the upper and lower limits of the range. and This ensures that the torque output can trigger "emergency deceleration" while preventing wheel lock-up.

[0183] The braking intent progresses from "slight → normal → emergency," corresponding to a gradual increase in the driver's "deceleration demand intensity." To achieve a precise match where "the stronger the intent, the more adaptable the braking torque output range is to high-intensity demands, and the wider the adjustment space covers extreme conditions," the "lower limit" and "upper limit" of the braking torque constraint range for each wheel need to increase synchronously, with the specific relationship as follows:

[0184] ;

[0185] ;

[0186] in, For the index of the wheel, , These correspond to the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0187] Based on the current driver braking intention category, obtain the corresponding initial constraint range of braking torque for each wheel.

[0188] Different braking intentions correspond to different driving scenario requirements (smooth braking for light braking, efficient braking for emergency braking). By matching dedicated torque constraint ranges to these three types of intentions, the braking torque output can be precisely matched to the scenario requirements. During light braking, the torque is limited to... Within this range, only a small braking force is output to avoid the vehicle "jerk" due to excessive torque. This range is suitable for scenarios that require smooth driving, such as following in congested traffic and slow deceleration. It can produce a perceptible slight deceleration without making passengers feel forward, thus maximizing driving comfort and smoothness.

[0189] During normal braking, the torque is limited to Within this range, it outputs "medium intensity and adjustable" braking force. This range is suitable for daily routine deceleration scenarios, such as deceleration at intersections, returning to position after overtaking, and speed limit adjustments on ring roads. At this time, the braking torque corresponds to a deceleration of 0.3g-0.8g. Passengers have a slight deceleration sensation but no discomfort from forward lurching. This ensures deceleration efficiency without affecting driving stability due to excessive torque, thus balancing efficiency and comfort in daily driving.

[0190] During emergency braking, the torque is limited to This range allows the system to output high-intensity braking force close to the tire's adhesion limit. This range is suitable for sudden hazardous scenarios, quickly increasing deceleration to 0.8g-1.2g to minimize braking distance and prioritize safety. Meanwhile, the lower limit of this range... The torque is set close to the upper limit of normal braking to ensure that the system can skip the low torque adjustment phase and quickly reach effective emergency braking force, avoiding missing the best braking opportunity due to response delay, and further ensuring safety in emergency scenarios.

[0191] S4. Based on the vehicle steering wheel angle, dynamically adjust the upper and lower limits of the initial constraint range of braking torque of each wheel, randomly generate multiple sets of wheel braking torque combinations within the adjusted constraint range to construct an initial population, and use the index prediction model trained based on wheel braking torque to predict the lateral acceleration and braking distance corresponding to each wheel braking torque combination.

[0192] Based on the above embodiments, the upper and lower limits of the braking torque constraint range of each wheel are dynamically adjusted according to the vehicle steering wheel angle. The specific logic is as follows:

[0193] When the steering wheel angle is less than the first preset angle threshold, the steering wheel angle is extremely small within this range, the vehicle's trajectory remains essentially straight, and the ground adhesion of the four wheels is evenly distributed during braking, resulting in a very low risk of lateral slippage. Therefore, the above state is defined as straight-line driving or slight steering. During straight-line driving or slight steering, there is no need to adjust the initial constraint range of the braking torque; maintaining the original range is sufficient—ensuring braking efficiency, i.e., the maximum torque can be output during emergency braking in a straight line, while avoiding braking delay caused by over-adjustment.

[0194] When the steering wheel angle is greater than or equal to the first preset angle threshold and less than or equal to the second preset angle threshold, the vehicle exhibits steering action within this range, but the steering angle is small, and the centrifugal force and lateral force of the vehicle body are within a "controllable range." Therefore, this state is defined as moderate steering. At this time, an "amplitude adjustment coefficient" is required. "Slightly lower the upper limit of torque—for example, when turning, the inner wheel's traction on the ground is slightly lower than that of the outer wheel, by adjusting the coefficient by the magnitude." By finely adjusting the torque of the inner wheel according to the steering angle, the braking effect is not affected, and the adhesion of the inner and outer wheels is balanced, avoiding understeer or oversteer caused by excessive torque on one side of the wheel.

[0195] When the steering wheel angle is greater than the second preset angle threshold, the steering angle is large within this range, the centrifugal force of the vehicle body is strong, and the difference in ground adhesion between the inner and outer wheels is significant (the outer wheels have greater adhesion, and the inner wheels have less). Therefore, the above state is defined as a sharp turn.

[0196] The methods for setting the first preset angle threshold and the second preset angle threshold are as follows:

[0197] The distribution of steering wheel angles during straight-line driving was statistically analyzed, and the "maximum angle for daily steering corrections" (e.g., when the driver makes minor steering adjustments, the maximum steering wheel rotation angle typically does not exceed 5°-8°) was taken as the baseline. This ensures that most "minor steering corrections" during straight-line driving are classified as "straight / slight steering";

[0198] The distribution of steering wheel angles by drivers in typical steering scenarios, such as turning at urban intersections and changing lanes on national highways, was statistically analyzed, and the 70%-80% quantile of steering angles in these scenarios was taken as the baseline. In sharp turns, such as sharp bends on rural roads or when avoiding obstacles, the turning angle is usually more than 30°-35°, thus clearly distinguishing it from regular turns.

[0199] Specifically, the upper and lower limits of the initial constraint range for the braking torque of each wheel are adjusted for different steering states, as follows:

[0200] When driving in a straight line or making a slight turn, the initial constraint range of the braking torque of each wheel remains unchanged;

[0201] In moderate steering situations, by statistically analyzing the steering wheel angle distribution of a large number of drivers in moderate steering scenarios (such as regular turns at urban intersections and smooth lane changes on national highways), the "average or median of the operating angle" in these scenarios is taken as the first steering wheel angle. :

[0202] ;

[0203] in, This is the first preset angle threshold, which is the dividing line between slight steering and moderate steering. The second preset angle threshold is the dividing value between moderate and severe steering.

[0204] The formula used to calculate the amplitude adjustment factor under moderate steering conditions is as follows:

[0205] ;

[0206] in, This is the amplitude adjustment coefficient under moderate steering conditions. The current steering wheel angle;

[0207] Wherein, if the current steering angle Just equal to ,but This indicates that there is no need to adjust the upper limit of torque; if The closer ,but The closer it is to 1, the larger the steering range, and the more significantly the upper limit of torque needs to be adjusted.

[0208] The formula used to calculate the upper limit of the initial constraint range for the adjusted braking torque of each wheel is as follows:

[0209] ;

[0210] in, For the adjusted number The upper limit of the initial constraint range for the braking torque of each wheel. The first preset for moderate steering conditions The adjustment range of the upper limit value for each wheel;

[0211] For the inner wheel, generally speaking, Will be set as 10% to 30%; for the outer wheels, generally speaking, Will be set as 0% to 10%.

[0212] Further explanation:

[0213] When a vehicle brakes lightly, each wheel has an initial constraint range for braking torque. ,in, This is the maximum braking torque that the wheel can withstand at this moment; this is the "initial reference" for adjustment.

[0214] When a vehicle makes a moderate turn (such as turning at an intersection or making a smooth lane change), it will "swing" outwards, resulting in:

[0215] When turning right, the outer wheel is pressed more firmly against the car body, increasing grip. It appears "conservative"; when turning right, the inner wheel is "lifted" more by the car body, resulting in weaker grip and reduced traction. It has become "dangerous". Therefore, adjustments must be made. The initial reference point used during slight braking can no longer be used.

[0216] First, based on the "changes in grip between the inner and outer wheels", pre-set the "adjustment direction and maximum adjustment range" for each wheel, that is... ;

[0217] Outer wheels: With increased grip, their maximum torque does not need to be lowered or only slightly lowered. It can be set close to 0, and the adjusted upper limit remains basically unchanged;

[0218] The inner wheel: With weakened traction, if the original maximum torque is applied, it's easy for the braking force to exceed the traction, causing slippage and loss of vehicle control. Setting it to a negative value means that the larger the steering angle, the more the upper limit of torque is lowered.

[0219] The degree of moderate steering varies (e.g., "a slight turn" versus "quickly turning to the limit of moderate steering"), and the degree of change in grip also differs. Therefore, a "grip adjustment factor" is needed to control the actual amount of adjustment.

[0220] The smoother the steering (closer to a normal turning angle), the smaller the change in grip, the closer the adjustment coefficient is to 0, and the weaker the adjustment force.

[0221] The more aggressive the steering (closer to the limit of moderate steering), the greater the change in grip, the closer the adjustment coefficient is to 1, and the stronger the adjustment force.

[0222] Therefore, the upper limit of the initial constraint range of the braking torque after adjustment for each wheel is equal to the upper limit of the initial reference value, plus the product of the preset upper limit adjustment range and the range adjustment coefficient for that wheel.

[0223] During sharp turns, through real-vehicle extreme condition tests, the critical steering angle at which the vehicle is "close to losing control but still controllable" was determined. At this angle, the centrifugal force of the vehicle body is relatively large, and the difference in adhesion between the inner and outer wheels is close to the limit. This angle serves as the "upper limit reference for safe steering" during sharp turns and is set as the second steering wheel angle. :

[0224] ;

[0225] The formula used to calculate the amplitude adjustment coefficient during a sharp turn is as follows:

[0226] ;

[0227] in, This is the amplitude adjustment coefficient during sharp turns. This refers to the maximum permissible value of the steering wheel angle, such as the common 540°-720° for passenger cars;

[0228] If the current steering angle Just equal to ,but The adjustment range is the smallest; if The closer ,but The closer it is to 1, the greater the need to lower the upper limit of torque to ensure safety.

[0229] When making a sharp turn, if the original maximum torque is maintained, the inner wheels are prone to locking up due to insufficient traction, leading to fishtailing or loss of control. Therefore, it is necessary to "further lower the maximum torque" based on the moderate steering adjustment, by... The adjustment range is quantified based on the difference between the steering angle and the maximum angle to ensure that even during sharp cornering and braking, the torque of each wheel will not exceed the current adhesion limit, thus prioritizing vehicle stability.

[0230] The formula used to calculate the upper limit of the initial constraint range of the wheel braking torque under sharp turning conditions is as follows:

[0231] ;

[0232] in, For the adjusted number The upper limit of the initial constraint range for the braking torque of each wheel. The preset first step for sharp turns The adjustment range for the upper limit of each wheel is a negative value;

[0233] For the inner wheel, generally speaking, Will be set as 10% to 20%; for the outer wheels, generally speaking, Will be set as 5% to 15%.

[0234] Further explanation:

[0235] When a vehicle makes a sharp turn, the centrifugal force increases significantly, the vertical load difference between the inner and outer wheels is much greater than that during moderate steering, and the difference in adhesion is even more pronounced. The size is determined by whether the wheel is on the "inner" or "outer" side, and the adjustment range must be greater than [a certain value]. .

[0236] The outer wheel (right wheel when turning right): While its traction is still greater than the inner wheel, the vehicle experiences significant body roll. Excessive torque will exacerbate the roll and increase the risk of loss of control. It is a moderately negative value;

[0237] The inner wheel (such as the left wheel when turning right): traction decreases drastically, making it a key risk point for fishtailing and loss of control. It is a negative value with a larger absolute value.

[0238] In the adjusted number The upper limit of the initial constraint range for the braking torque of each wheel is equal to... When, then the adjusted number The initial constraint range for the braking torque of each wheel is: .

[0239] Further explanation:

[0240] Sharp turns are a more extreme situation than moderate turns, where the difference between centrifugal force and wheel adhesion is much greater. Therefore, torque adjustments for sharp turns do not start directly from the initial reference of straight-line conditions, but rather from the "upper limit after moderate turn adjustments."

[0241] When a vehicle enters a sharp turn, the most critical change is that the centrifugal force is greater than during a moderate turn. The vehicle body will tilt violently to the outside, and the vertical load on the inner wheel may drop to less than 30% of the static load, or even close to "lifting off the ground"—this means that the adhesion of the inner wheel decreases sharply, and even a little braking force may exceed the limit and lock up.

[0242] Although the outer wheels are "pressed tighter", the body roll is already close to the suspension's load limit. If the braking torque is too large, it will further aggravate the roll and even risk rollover. At this point, the torque adjustment range during moderate steering is not enough - it is necessary to "press down" further on this basis to control the torque within the adhesion limit.

[0243] The moderate steering setting has already undergone adjustments to address the "difference in grip during steering" (inner side adjusted downwards, outer side basically unchanged), and represents the "basic safety value" for steering conditions; sharp turns... Starting from this point, it's like "adding an extra layer of insurance based on the extreme risks of sharp turns, on top of the already adapted safety line for steering," making the adjustments more consistent and precise.

[0244] The risks of sharp turns are not constant; rather, they increase progressively with the increase in the steering angle. Therefore, it is necessary to... , This coordination enables a dynamic match between the "level of danger" and the "range of adjustment."

[0245] With a critical steering angle that is "close to being out of control but still controllable" For reference, use "current steering angle" and The "deviation" converts the "level of danger" into a quantitative coefficient of 0-1:

[0246] when near : The deviation is small, A value close to 0 indicates a low level of danger, requiring only a slight tightening of the torque on a moderate basis to avoid excessively sacrificing braking efficiency;

[0247] when near : The deviation is large, A value close to 1 indicates a high level of danger, requiring immediate action. Even if the braking distance increases, the maximum reduction should be made to prioritize vehicle stability.

[0248] In summary, the preset first step during a sharp turn... Adjustment range of the upper limit value for each wheel Amplitude adjustment coefficient during sharp turns Multiply by this to get the "actual adjustment amount during sharp turns"; then add this adjustment amount to the adjusted value for moderate turns. Upper limit of the initial constraint range for the braking torque of each wheel The adjusted position under sharp turn conditions is obtained. Upper limit of the initial constraint range for the braking torque of each wheel .

[0249] Based on the current steering state, obtain the upper and lower limits of the corresponding adjusted braking torque constraint range for each wheel.

[0250] Further explanation:

[0251] By adjusting the braking torque constraint range of each wheel in stages according to the steering state (straight line / slight steering → moderate steering → sharp turn), the upper limit of the torque is precisely matched with the wheel adhesion limit under the current steering condition.

[0252] Maintain the original range when driving in a straight line to ensure maximum braking efficiency (such as outputting maximum torque during emergency braking).

[0253] When steering in a moderate manner, fine-tune the upper limit of the torque on the inner wheel to balance the adhesion between the inner and outer sides and avoid excessive torque on one side, which could lead to understeer or oversteer.

[0254] When making sharp turns, the upper limit of the torque on the inner wheel is significantly reduced to adapt to the drastically reduced traction, preventing wheel lock-up or vehicle fishtailing from the source.

[0255] This dynamic adjustment of "higher steering risk → stricter torque constraint" directly limits the braking torque within a safe range, ensuring braking stability under different steering scenarios from the ground up.

[0256] In addition, the adjusted constraint interval provides a precise and reasonable boundary for the random generation of "multiple wheel braking torque combinations":

[0257] Avoid generating invalid combinations that exceed the safety range of the current steering scenario (e.g., avoid generating combinations with excessive torque on the inner wheels during sharp turns), and reduce the algorithm's invalid calculations of "unsafe solutions";

[0258] By focusing on the torque range that "meets both the current adhesion limit and the braking requirements", the initial population is made closer to the search space of the optimal solution, shortening the convergence time of the genetic algorithm and improving the solution efficiency.

[0259] In short, dynamically adjusted constraint intervals are equivalent to "defining a safe and effective search range" for the algorithm, avoiding blind searches and wasting computing power.

[0260] Furthermore, by adjusting the torque constraints of the inner and outer wheels differently (e.g., a more significant downward adjustment on the inner wheel during steering), it is possible to ensure overall braking performance (e.g., achieving the required braking distance) while also specifically balancing the vehicle's posture during steering.

[0261] It refines the constraints of different steering angles and different wheels (inner and outer) to adapt to the dynamic force state of the vehicle during steering.

[0262] By combining the prediction model of lateral acceleration (reflecting roll stability) and braking distance (reflecting braking efficiency), a "dual-objective evaluation benchmark" is provided for subsequent optimization, ensuring that the final output torque combination minimizes the risk of sideslip without sacrificing braking distance.

[0263] Based on the above embodiments, multiple sets of wheel braking torque combinations are randomly generated within the adjusted constraint interval to construct an initial population. The specific logic is as follows:

[0264] Within the constrained range of the adjusted braking torque of each wheel, the braking torques of the left front wheel, right front wheel, left rear wheel, and right rear wheel are randomly generated to form an initial population. This initial population is then labeled as... , , The first in the initial population Individual, This is the index of an individual in the initial population, and , The initial population size. ,in, The first The braking torque of the left front wheel, right front wheel, left rear wheel, and right rear wheel of each individual.

[0265] Based on the above embodiments, the indicator prediction model is constructed using a deep learning network based on a multilayer perceptron. The deep neural network of the multilayer perceptron includes an input layer, a first hidden layer, a second hidden layer, and an output layer. The first hidden layer and the second hidden layer each have at least two neurons and both use ReLU (linear rectified unit) as the activation function.

[0266] The braking torque combination of each wheel, the corresponding lateral acceleration, and the braking distance are divided into training set, validation set, and test set, with a common ratio of 7:2:1. The training set is used to fit the mapping relationship between input and output; the validation set is used to adjust the learning rate and the number of hidden layer neurons during training to prevent overfitting; and the test set is used to evaluate the generalization ability of the model after training.

[0267] The structure of a deep learning network with a multilayer perceptron includes:

[0268] Input layer: Used to receive the combination of braking torques of each wheel;

[0269] The first hidden layer has 32 neurons and uses ReLU as the activation function.

[0270] The second hidden layer has 16 neurons and also uses the ReLU activation function;

[0271] Output layer: It has 2 neurons, which are used to output lateral acceleration and braking distance respectively. The output layer does not require an activation function and directly outputs continuous values.

[0272] The current task is "input 4 braking torques → output 2 vehicle indicators", which belongs to the "low-dimensional input, simple mapping relationship" task. The shallow network (1-2 hidden layers, 32-64 neurons) has few parameters, which can learn the non-linear relationship between braking torque and indicators, and ensure fast training and strong generalization ability, making it suitable for this kind of simple regression task.

[0273] The process of training the indicator prediction model is as follows:

[0274] Using the combination of braking torques of each wheel as input and the corresponding lateral acceleration and braking distance as output labels, a feature association model is trained, using mean squared error as the loss function. When the mean squared error is within a certain range... When the value is within the specified range, stop training the indicator prediction model.

[0275] S5. Process lateral acceleration and braking distance to generate comprehensive evaluation coefficients. With minimizing the comprehensive evaluation coefficients as the objective function, use an optimization algorithm to iteratively optimize the wheel braking torque combination to obtain the optimal wheel braking torque combination.

[0276] Table 1. Changes in the comprehensive evaluation coefficient with normalized lateral acceleration and braking distance

[0277]

[0278] According to Table 1, when the normalized lateral acceleration deviation value changes from 0 to 0.96 and the normalized braking distance changes from 0.07 to 0.69, the comprehensive evaluation coefficient changes from 0.04 to 0.74, showing a monotonically increasing trend throughout. Therefore, the normalized lateral acceleration deviation value, the normalized braking distance, and the comprehensive evaluation coefficient are all positively correlated. The normalization process did not change the correlation logic of the original indicators.

[0279] according to Figure 2 It was found that the data points (black squares) were closely distributed around the fitted straight line. As the "normalized lateral acceleration deviation value" increased from 0 to 0.96, the "comprehensive evaluation coefficient" increased monotonically from 0.04 to 0.74, and the two showed completely consistent trends.

[0280] The slope of the fitted line is positive and relatively steep, indicating that for every unit increase in the lateral acceleration deviation, the comprehensive evaluation coefficient will have a significant positive increase, reflecting a strong linear correlation that "the larger the lateral acceleration deviation, the larger the comprehensive evaluation coefficient".

[0281] according to Figure 3 It was found that the data points (black squares) also closely matched the fitted straight line. When the "normalized braking distance" increased from 0.07 to 0.69, the "comprehensive evaluation coefficient" increased from 0.04 to 0.74 in sync, and the trend of change was completely synchronized.

[0282] The slope of the fitted straight line is positive, indicating that the comprehensive evaluation coefficient will increase significantly for every unit increase in braking distance, reflecting the linear relationship that "the longer the braking distance, the larger the comprehensive evaluation coefficient".

[0283] Based on the above embodiments, the lateral acceleration and braking distance are normalized. A comprehensive evaluation coefficient is generated using the normalized lateral acceleration and braking distance, based on the following formula:

[0284] ;

[0285] ;

[0286] in, For the first The comprehensive evaluation coefficient of each individual wheel braking torque combination is used to evaluate the stability and efficiency of the wheel braking torque combination by combining two index parameters: lateral acceleration deviation and braking distance. The smaller the comprehensive evaluation coefficient, the higher the stability and efficiency of the wheel braking torque combination.

[0287] In the formula, For the first The lateral acceleration deviation value of each individual. For the first Braking distance of each individual;

[0288] Based on the above, it should be noted that:

[0289] Lateral acceleration describes a vehicle's dynamic performance during cornering or lateral movement. It is relative to the maximum lateral acceleration a vehicle can withstand. Every vehicle has a safe range of lateral acceleration. If the lateral acceleration is too high, the vehicle may lose traction, leading to skidding or rollover; if it is too low, the vehicle may not be able to effectively handle cornering, affecting the driving experience. Therefore, there is a suitable range for lateral acceleration, and it is not always better to be as low as possible.

[0290] Braking distance is an absolute measure that directly reflects the distance a vehicle needs to travel from the start of braking to a complete stop. Generally speaking, the shorter the braking distance, the more efficient the braking system and the better the vehicle's safety.

[0291] The lateral acceleration deviation value refers to the deviation of the lateral acceleration from the appropriate range. When the lateral acceleration deviation value decreases, the matching degree between the lateral force and the adhesion force improves, and the vehicle's dynamic stability is enhanced. Therefore, the overall evaluation coefficient decreases.

[0292] Braking distance refers to the distance a vehicle travels from the start of braking to a complete stop. When the braking distance decreases, the matching degree between longitudinal braking force and ground adhesion improves, and braking energy conversion becomes more efficient. Therefore, the overall evaluation coefficient decreases.

[0293] Therefore, the comprehensive evaluation coefficient is positively correlated with the lateral acceleration deviation and braking distance.

[0294] The stability-efficiency evaluation process of the wheel braking torque combination is a complex vehicle dynamics system involving multi-parameter coupling, multi-scenario adaptation, and multi-objective balancing. Lateral acceleration deviation reflects the lateral dynamic stability state (roll / slip risk) during vehicle braking, while braking distance reflects the longitudinal braking energy conversion efficiency (deceleration and stopping capability). Although these two indicators are influenced by factors such as ground adhesion and vehicle attitude in vehicle dynamics, within a local evaluation window for specific steering states (such as straight lines, moderate steering, and sharp turns), the interaction patterns of lateral forces (centrifugal force and lateral adhesion) and longitudinal forces (braking force and longitudinal adhesion) can be considered independent performance contribution factors. Based on the principle of independent force action in vehicle dynamics (the influence of forces in different directions on the motion state can be analyzed separately and then superimposed), the influence of the two independent performance factors on the comprehensive performance of the braking torque combination can be considered as a linear superposition.

[0295] Furthermore, the lateral acceleration deviation and braking distance characterize the actual performance of the braking torque combination from two core dimensions: lateral stability and longitudinal efficiency. The former focuses on the vehicle posture safety during steering and braking, while the latter focuses on the distance efficiency of deceleration and stopping. The two have clear independence in terms of definition and physical meaning, and there is no overlap or logical confusion in dimensions. The stability-efficiency of the wheel braking torque combination can be evaluated by linear weighting.

[0296] In summary, the functional relationship between the comprehensive evaluation coefficient, the lateral acceleration deviation, and the braking distance is expressed in the above-mentioned functional form.

[0297] In the formula, For the first The lateral acceleration of each individual, This represents the lower limit of the suitable range for lateral acceleration, which is between 0.2g and 0.3g. This represents the upper limit of the suitable range for lateral acceleration, which is between 0.6g and 0.9g.

[0298] In the formula, , These are the weighting coefficients for the lateral acceleration deviation and the braking distance, respectively. , The specific value is determined by the Analytic Hierarchy Process (AHP), and the specific logic is as follows:

[0299] Two indicators, lateral acceleration deviation and braking distance, are labeled. The relative importance of each pair of indicators is determined using the nine-scale method, and a judgment matrix is ​​constructed. The index of lateral acceleration deviation is labeled as 1, and the index of braking distance is labeled as 2. for:

[0300] ;

[0301] in, , Both represent the index of the parameter, and , , indicating that the index is The parameter relative to the index is Regarding the parameters, their importance to the comprehensive evaluation coefficient is... The specific values ​​were determined by relevant experts using a 1-9 scoring method. Indicates that the index is The parameters compared to the index are Regarding the parameters, the comprehensive evaluation coefficient is extremely important. Indicates that the index is The parameters compared to the index are For parameters, the comprehensive evaluation coefficient is extremely unimportant;

[0302] Divide each element value in the judgment matrix by the sum of its columns to obtain a normalized judgment matrix. Calculate the mean of the element values ​​in each row of the normalized judgment matrix. Use the mean of the element values ​​in the first row as the scaling factor for the lateral acceleration deviation value and the mean of the element values ​​in the second row as the scaling factor for the braking distance. With the constraint that the sum of the scaled values ​​equals 1, scale the two scaling factors proportionally and use the scaled values ​​as the weights of the corresponding parameters.

[0303] Based on the above embodiments, with the minimization of the comprehensive evaluation coefficient as the objective function, an optimization algorithm is used to iteratively optimize the wheel braking torque combination to obtain the optimal wheel braking torque combination. The specific logic is as follows:

[0304] Within the constraint range of braking torques for the left front wheel, right front wheel, left rear wheel, and right rear wheel, iterative optimization is performed on individuals in the initial population. Specifically, this involves adjusting the comprehensive evaluation coefficient... Sort by size from largest to smallest, and select the comprehensive evaluation coefficient. Individuals ranked at the top are considered parents; "top" refers to the individuals in the sorted set with the highest overall evaluation coefficient. The top 50% of individuals are used in a genetic algorithm to crossover and mutation operations on their parent genes, generating new individuals. A predictive model is used to predict the lateral acceleration and braking distance of these new individuals, and the corresponding comprehensive evaluation coefficient is calculated. These new individuals are then merged with their parent genes to form the next generation population. The selection, crossover, and mutation operations are repeated. When convergence is achieved, the individual with the minimum comprehensive evaluation coefficient is selected as the optimal wheel braking torque combination, denoted as... ,in, These are the braking torques of the left front wheel, right front wheel, left rear wheel, and right rear wheel in the optimal wheel braking torque combination.

[0305] S6. Distribute the braking torque of each wheel according to the optimal wheel braking torque combination.

[0306] Based on the optimal braking torque combination, a total braking torque is set, the specific braking torque of each wheel is calculated, and the total braking torque is distributed to each wheel proportionally. Generally, the braking torque of the front wheels will be higher than that of the rear wheels, especially during high-speed braking. A common ratio is that 60%-70% of the braking torque is distributed to the front wheels and 30%-40% to the rear wheels.

[0307] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0308] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented in software, the above embodiments can be implemented, in whole or in part, as a computer program product. Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented by software, electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution.

[0309] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment, depending on actual needs.

[0310] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. An EMB brake system failure vehicle stability control method, characterized by, The specific steps include: S1. Obtain the historical data of each wheel braking torque and vehicle dynamic parameters in normal state and failure state, extract the time sequence characteristics of the historical data, construct a multi-dimensional feature vector, and form a training sample; S2. Use the state discrimination model trained based on the training sample to discriminate the multi-dimensional feature vector of the current time period, and output the corresponding EMB braking system working state, which is normal state or failure state; S3. When it is determined to be normal state, the braking torque of each wheel is distributed according to the original distribution rule; when it is determined to be failure state, the current vehicle pedal parameter is collected, the corresponding driver braking intention category is identified, and the corresponding initial constraint interval of each wheel braking torque is matched; S4. Dynamically adjust the upper and lower limits of the initial constraint interval of each wheel braking torque based on the vehicle steering wheel angle, randomly generate multiple wheel braking torque combinations in the adjusted constraint interval to construct an initial population, and use the index prediction model trained based on the wheel braking torque to predict the lateral acceleration and braking distance corresponding to each wheel braking torque combination; S5. Process the lateral acceleration and braking distance to generate a comprehensive evaluation coefficient, take the minimization of the comprehensive evaluation coefficient as the objective function, and use an optimization algorithm to iteratively optimize the wheel braking torque combination to obtain the optimal wheel braking torque combination; S6. Distribute the braking torque of each wheel according to the optimal wheel braking torque combination.

2. The EMB brake system failure vehicle stability control method according to claim 1, characterized by, The vehicle dynamic parameters include vehicle speed and vehicle yaw rate; the pedal parameters include brake pedal stroke, brake pedal force and brake pedal speed.

3. The EMB brake system failure vehicle stability control method according to claim 1, characterized by, The time sequence characteristics of the historical data are extracted, and a multi-dimensional feature vector is constructed, with the specific logic as follows: The collected historical data includes the working state of the EMB braking system at each historical time; For each historical time, a time period is pushed forward, and the time sequence characteristics of each wheel braking torque and vehicle dynamic parameters are collected in the time period to form a multi-dimensional feature vector; A mapping relationship between the multi-dimensional feature vector of each time period and the working state of the EMB braking system at the corresponding historical time is established.

4. The EMB brake system failure vehicle stability control method according to claim 1, characterized by, The specific logic of distributing the braking torque of each wheel according to the original distribution rule is as follows: According to the driver's braking request, the total braking force required is calculated in combination with the current vehicle speed; The vertical load of each wheel and the friction coefficient between the tire and the road are obtained, and the maximum braking force that each wheel can withstand is calculated; According to the preset front and rear axle and left and right wheel distribution ratio, the total braking force is distributed in combination with the maximum braking force that each wheel can withstand, and the corresponding braking torque is applied to each wheel braking actuator in combination with the tire radius.

5. The EMB brake system failure vehicle stability control method according to claim 2, characterized by, The driver braking intention category includes slight braking, normal braking and emergency braking; The first preset threshold values of the brake pedal stroke, the brake pedal force, and the brake pedal speed are respectively defined as , , , and the second preset threshold values of the brake pedal stroke, the brake pedal force, and the brake pedal speed are respectively defined as , , ; If any of the above pedal parameters of the current vehicle exceeds the corresponding second preset threshold, it is emergency braking; If any of the above pedal parameters of the current vehicle does not exceed the corresponding second preset threshold, and any pedal parameter is between the first preset threshold and the second preset threshold, it is normal braking; If any of the above pedal parameters of the current vehicle does not exceed the corresponding first preset threshold, it is slight braking.

6. The EMB brake system failure vehicle stability control method according to claim 5, characterized by, According to the current driver braking intention category, the corresponding wheel braking torque initial constraint interval is matched, and the specific logic is as follows: Defining initial constraint intervals for each wheel braking torque ; wherein, are the initial constraint intervals for the braking torque of the first wheel, are the initial constraint intervals for the braking torque of the first wheel, wherein: ； ； wherein, respectively the lower limit value of the initial constraint interval of the braking torque of the first wheel under light braking, normal braking, emergency braking, respectively the upper limit value of the initial constraint interval of the braking torque of the first wheel under light braking, normal braking, emergency braking, Wherein, for any wheel, the following relationship is satisfied: ； ； wherein, is an index of the wheel, , corresponding to the left front wheel, the right front wheel, the left rear wheel, the right rear wheel, respectively; According to the current driver braking intention category, the corresponding wheel braking torque initial constraint interval is obtained.

7. The EMB brake system failure vehicle stability control method according to claim 6, characterized by, Based on the vehicle steering wheel angle, the upper and lower limits of the wheel braking torque constraint interval are dynamically adjusted, and the specific logic is as follows: When the steering wheel angle is less than the first preset angle threshold, it is defined as straight driving or slight turning; When the steering wheel angle is greater than or equal to the first preset angle threshold and less than or equal to the second preset angle threshold, it is defined as moderate turning; When the steering wheel angle is greater than the second preset angle threshold, it is defined as sharp turning; Wherein, for different steering states, the upper and lower limits of the wheel braking torque initial constraint interval are adjusted, and the specific logic is as follows: In the straight driving or slight turning state, the wheel braking torque initial constraint interval remains unchanged; In the moderate steering state, the first steering wheel angle is set : ； wherein, is a first preset angle threshold, is a second preset angle threshold; The amplitude adjustment coefficient in the moderate turning state is calculated, and the formula is as follows: ； wherein, is the amplitude adjustment factor in the moderate steering state, is the current steering wheel angle; The adjusted upper limit of the wheel braking torque initial constraint interval is calculated, and the formula is as follows: ； wherein, is the upper limit value of the adjusted initial constraint interval of the wheel braking torque, is the preset adjustment range of the upper limit value of the wheel in the moderate steering state. In the state of sharp turn, the second steering wheel angle is set : ； The amplitude adjustment coefficient in the sharp turning state is calculated, and the formula is as follows: ； wherein is a magnitude adjustment factor for a sharp turn, is a maximum allowable value of the steering wheel angle; In the sharp turning state, the wheel braking torque initial constraint interval upper limit value in the moderate turning state is further adjusted, and the wheel braking torque initial constraint interval upper limit value in the sharp turning state is obtained, and the formula is as follows: ； wherein, is the upper limit value of the adjusted initial constraint interval of the wheel braking torque, is the upper limit value of the adjusted initial constraint interval of the wheel braking torque in the sharp turn state. According to the current steering state, the upper and lower limits of the adjusted wheel braking torque constraint interval are obtained.

8. The EMB brake system failure vehicle stability control method according to claim 7, characterized by, Randomly generate multiple groups of wheel braking torque combinations in the adjusted constraint interval to construct an initial population, and the specific logic is as follows: Within the constrained range of the adjusted braking torque of each wheel, the braking torques of the left front wheel, right front wheel, left rear wheel, and right rear wheel are randomly generated to form an initial population. This initial population is then labeled as... , , The first in the initial population Individual, This is the index of an individual in the initial population, and , The initial population size. ,in, The first The braking torque of the left front wheel, right front wheel, left rear wheel, and right rear wheel of each individual.

9. The EMB brake system failure vehicle stability control method according to claim 8, characterized by, The lateral acceleration and braking distance are normalized, and the normalized lateral acceleration and braking distance are used to generate a comprehensive evaluation coefficient, and the formula is as follows: ； ； wherein, is the comprehensive evaluation coefficient of the first individual, is the lateral acceleration deviation value of the first individual, is the braking distance of the first individual; wherein is the lateral acceleration of the is the lateral acceleration of the is the lower limit value of the lateral acceleration suitable interval, is the upper limit value of the lateral acceleration suitable interval; In the formula, , respectively are the lateral acceleration deviation value, the weight coefficient of the braking distance, , The specific values of are determined by the analytic hierarchy process.

10. The EMB brake system failure vehicle stability control method according to claim 9, characterized by, Taking the minimization of the comprehensive evaluation coefficient as the objective function, an optimization algorithm is used to iteratively optimize the wheel braking torque combination to obtain the optimal wheel braking torque combination, and the specific logic is as follows: In the constraint interval of the left front wheel, right front wheel, left rear wheel and right rear wheel braking torque, the individuals in the initial population are iteratively optimized, specifically the comprehensive evaluation coefficient Sort from large to small, select the comprehensive evaluation coefficient The individuals in the front are selected as parents, and the gene coding of the parent individuals is exchanged and combined and mutated through the crossover and mutation operations in the genetic algorithm to generate new individuals. The lateral acceleration and braking distance of the newly generated individuals are predicted using the index prediction model, and the corresponding comprehensive evaluation coefficient is calculated. The newly generated individuals and the parent individuals are combined as the next generation population, and the selection, crossover and mutation operations are repeated. When the convergence condition is met, the individual corresponding to the minimum comprehensive evaluation coefficient is selected as the optimal wheel braking torque combination, denoted as Wherein, The left front wheel, right front wheel, left rear wheel and right rear wheel braking torque in the optimal wheel braking torque combination, respectively.

Images