Pedal characteristic design method for brake-by-wire system based on online identification of brake individuality

Abstract

The application provides a brake pedal characteristic design method of a brake-by-wire system based on online identification of braking individuality, comprising the following steps: extracting braking characteristic data, dividing driver braking individuality types, and designing individualized assist braking characteristic MAP curves; in the driving process of the driver, a BP network model is applied to online identification of the driver braking individuality, a driver braking individuality type signal is decided by Bayes fusion, and the signal is stored; an electronic assist braking system ECU receives the signal and matches the individualized assist braking characteristic MAP curves, individualized braking assist functions are realized through a vehicle execution module, and individualized braking pedal characteristics are provided; the application can further refine the individualized pedal characteristics of the driver, the automatic identification and determination of the pedal characteristics do not require the driver to manually select a braking mode, the individualized braking pedal feeling requirements are better met, the time and difficulty of the automobile manufacturer in calibrating the electronic assist braking system are reduced, and the intelligent human-vehicle interaction level is comprehensively improved.

Description

Technical Field

[0001] This invention relates to the field of automotive control technology, and in particular to a method for designing pedal characteristics of a brake-by-wire system based on online identification of braking characteristics. Background Technology

[0002] Currently, with the development of automotive electronics and intelligence, electronic power-assisted braking systems, with their advantages such as electronic unit control, dual redundancy backup, and adjustable assist motor torque, have become the main development trend of electric vehicle braking systems. At present, people have increasingly higher demands for automotive comfort and personalization. Because different types of drivers have significantly different braking behaviors, a mismatch between the response of electronic power-assisted braking systems and driver individuality can reduce driver satisfaction and acceptance of the braking system.

[0003] Generally, the assist performance of electronic power-assisted braking systems is integrated into the vehicle after parameters are calibrated by OEM engineers and provided to consumers. Different models mostly use the same set of calibration parameters, therefore, it cannot 100% meet the driving and control needs of different drivers. With the development of market demand, mode-switching functions have gradually been developed in high-end vehicles, generally divided into three modes: Comfort, Normal, and Sport, which are switched by the driver manually selecting a mode button. However, this method of adding buttons not only increases development costs and the complexity of the human-machine interface, but also fails to automatically adjust and switch based on real-time driving data. Furthermore, the three mode types are relatively limited and cannot cover the diverse braking personalities of drivers in today's society. Therefore, integrating online identification of driver braking personality types and further refining driver braking personality types into Conservative Type I, Conservative Type II, Normal, Aggressive Type I, and Aggressive Type II, and parametrically designing personalized power-assisted braking characteristic MAP curves to provide a brake pedal feel adapted to the driver's braking personality type, is a key technology for the future development of intelligent vehicle brake-by-wire systems. Summary of the Invention

[0004] Purpose of the invention: The technical problem to be solved by the present invention is to address the shortcomings of the prior art by providing a method for designing pedal characteristics of a brake-by-wire system based on online identification of braking personality. This method can automatically identify the braking personality type of different drivers online and adjust the personalized braking assist control strategy according to different drivers' braking characteristic data, thereby meeting the driving operation needs of different types of drivers, ensuring driving safety, and providing drivers with a better driving experience.

[0005] This invention is based on an electronic power-assisted braking system and includes a braking characteristic data acquisition module, a driver braking personality online identification module, a personalized braking assist control module, and a vehicle execution module.

[0006] The present invention specifically includes the following steps:

[0007] Step 1: Set up a preliminary database to store the full-condition driving braking dataset of the driver driving the target test vehicle and the online identification results of the driver's braking personality type; in the vehicle controller, use deviation standardization to perform dimensionality reduction on the driving braking data, and use principal component analysis to extract braking feature data that can characterize the driver's braking personality to form a training sample set S.

[0008] Step 2: Apply the K-Means algorithm to the training sample set S to perform cluster analysis on the driver's braking personality and classify the driver's braking personality type;

[0009] Step 3: Based on the driver's braking personality type, design a personalized MAP curve for the power-assisted braking characteristic using parametric design to provide different brake pedal characteristics;

[0010] Step 4: During the driver's driving process, the driver's braking personality type is identified online using a BP neural network model based on the vehicle braking data. Combined with the driver braking personality type identification results stored in the previous database, the Bayesian fusion decision method is used to determine the current driver's braking personality type signal with credibility and store it in the previous database.

[0011] Step 5: The Electronic Assisted Braking System (ECU) receives the current driver's braking personality type signal, matches the personalized assisted braking characteristic MAP curve, realizes personalized assisted braking function, and provides brake pedal characteristics corresponding to the driver's braking personality type.

[0012] In step 1, the full-condition driving braking data set of the driver-driven target test vehicle includes the following data: driving braking data collected by radar and camera sensors, including the longitudinal speed of the preceding vehicle and the relative distance between the driver and the preceding vehicle; driving braking data collected based on vehicle CAN bus communication, including the driver's longitudinal speed, braking deceleration, brake pedal travel, brake master cylinder pressure, and battery status; and extracted braking feature data that characterizes the driver's braking personality, including: driver braking deceleration k1, brake master cylinder pressure k2, longitudinal relative speed between the driver and the preceding vehicle k3, and relative distance between the driver and the preceding vehicle k4. A training sample set S is constructed using the braking feature data. The training sample set S contains m samples, where m is a natural number. Each sample contains a set of braking feature data characterizing the driver's braking personality. i Let s represent the i-th data point in the training sample S, where i = 1, 2, ..., m. i =(k i1 ,k i2 ,k i3 ,k i4 ), k i1 ,k i2 ,k i3 ,k i4These represent the vehicle's braking deceleration, master cylinder pressure, longitudinal relative speed between the vehicle and the vehicle in front, and relative distance between the vehicle and the vehicle in front, respectively, in the i-th data in the training sample S.

[0013] In step 2, the K-Means algorithm is applied to cluster the m samples in the training sample set S into 5 classes, and then the following steps are performed:

[0014] Step 2-1: Randomly select five cluster centers s from the training sample set S. r r takes values ​​from 1 to 5;

[0015] Step 2-2: The remaining data samples, excluding the five cluster centers, are assigned to the data clusters formed by one of the five cluster centers.

[0016] Steps 2-3: Recalculate the cluster centers s of the five data clusters based on the mean principle. r ;

[0017] Steps 2-4: Calculate the measure function D:

[0018]

[0019] Where m represents the total number of training samples; r represents the r-th personality label, and the values ​​of r, 1, 2, 3, 4, and 5 correspond to conservative type I, conservative type II, normal type, radical type I, and radical type II, respectively.

[0020] Step 2-5: If the measure function D converges, the cluster analysis ends; if the measure function D does not converge, return to step 2-2 and continue executing steps 2-2 to 2-4 until D converges, at which point the cluster analysis ends.

[0021] After the cluster analysis, all data samples were assigned a personality label. There are five personality labels: Conservative Type I, Conservative Type II, Normal Type, Radical Type I, and Radical Type II.

[0022] In step 3, five personalized assisted braking characteristic MAP curves are parametrically designed, including safety, comfort, normal, sport, and adventure assisted braking characteristic curves, to achieve different brake pedal characteristics and provide different types of brake pedal feel.

[0023] In step 3, the personalized power-assisted braking characteristic MAP curve represents different types of braking characteristics by setting the following parameters of the electronic power-assisted braking system: linear stage assist ratio, jump increment, and inflection point master cylinder pressure. The linear stage assist ratio is adjusted by changing the displacement difference between the primary and secondary surfaces of the feedback disc during the assist process. Different assist ratios are obtained by setting different displacement differences between the primary and secondary surfaces corresponding to different input push rod displacements. The jump increment is adjusted by regulating the correspondence between the displacement difference between the primary and secondary surfaces of the feedback disc and the input push rod displacement during the jump increment stage. For the same input push rod displacement, the larger the displacement difference between the primary and secondary surfaces of the feedback disc, the larger the corresponding jump increment. The assist characteristic curve of a traditional vacuum booster is selected as the conventional assist characteristic curve, with a linear stage assist ratio of 7.6, a jump increment of 16 bar, and an inflection point master cylinder pressure of 90 bar. Using the conventional assist characteristic curve as the design benchmark, the parameter values ​​are adjusted upwards and downwards to correspond to the safety, comfort, sport, and adventure assist characteristic curves, respectively.

[0024] In step 3, the safety-type power assist characteristic curve shows the largest jump increase and the largest assist in the linear stage, which is suitable for conservative type I drivers;

[0025] The comfort-type power steering characteristic curve has a larger jump increase and a larger linear stage of power assistance, making it suitable for conservative Type II drivers.

[0026] The standard power assist characteristic curve has a boost and linear phase power assist ratio that falls between comfort and sport, making it suitable for average drivers.

[0027] The sport-type power assist curve has a smaller jump increase and a smaller linear stage of power assist, making it suitable for aggressive Type I drivers.

[0028] The adventure-type boost curve has the smallest jump increase and the smallest linear boost, making it suitable for aggressive Type II drivers.

[0029] Step 4 includes:

[0030] The driver braking personality type y identified in the BP neural network during the nth driving process n And the set of driver braking personality types {y1y2y3…y1} stored in the previous database from the first n-1 online identifications. n-1 Synthesize a new data cluster Y n ={y1y2y3…y n-1 y n Then, a Bayesian fusion decision is performed to ultimately determine the driver's braking personality type x for that particular incident. n x n The values ​​from 1 to r correspond to r types of driver braking personality, with r = 5; additionally, r = 0 indicates a type that is difficult to distinguish.

[0031] Based on Bayes' theorem:

[0032] Where P represents the probability of the event; P(x) n |y n ) represents the driver braking personality type identified by the nth BP neural network, y. n Under the premise of fusion decision-making, the driver braking personality type is determined to be x. n The posterior probability; P(x) n |y n ,Y n-1 ) represents the driver braking personality type identified by the nth BP neural network, y. n Under the condition of fusion decision determination, the driver braking personality type is x n And the set of braking personality types of the drivers identified online in the previous n-1 times is Y. n-1 The probability of simultaneous occurrence; P(y) n |x n ,Y n-1 Given a driver braking personality type of x n Under the assumption that y occurs n And the set of braking personality types of the drivers identified online in the previous n-1 times is Y. n-1 The probability of both occurring simultaneously, P(y) n |Y n-1 Y represents the set of driver braking personality types identified in the first n-1 online tests. n-1 Under the premise of y n The probability is set as follows, based on the characteristics of the driver braking personality type sample:

[0033] The classification results are independent: P(y) n |x n ,Y n-1 )=P(y n |x n );

[0034] Driver's braking personality type x n Conforms to a Markov process: The classification results are independent of each other, and the driver's braking personality type is x n Substituting the results conforming to the MarKov process into Bayes' theorem, we get: Where P(y) n |x n () indicates that, given a driver braking personality type of x n Observing y under the assumption of class n The probability of P(x); due to P(x) n =i)(i=0,1,…,K) are unknown, assumed to be equal probabilities, P(x n=i) = 1 / (K+1), therefore, Among them, conf(x n The driver braking personality type provided by the classifier x n Confidence level;

[0035] P(x n |x n-1 () indicates the driver's braking personality type x n-1 The driver's braking personality type is x n The probability, Introducing manipulation tolerance γ and style tolerance ε, the final expression is: `max` retrieves the maximum value, and `min` retrieves the minimum value.

[0036] Set 0 ≤ ε ≤ 0.5 and 0 ≤ γ ≤ K; P(x n-1 |Y n-1 The result of the previous fusion decision calculation is given, with the initial condition that all style types have equal chances:

[0037] For x n =0,1,2,…K, calculate the posterior probability P(x) n |Y n ), thus obtaining the output vector [P(0|Y n )P(1|Y n )…P(K|Y n )], corresponding to the maximum P(x n |Y n The category is the current driver braking personality type c. n =argmaxP(x n |Y n ), where argmax(f(x)) is a function, find the independent variable x that maximizes the value of the function f(x) inside the parentheses of argmax(), argmaxP(x n |Y n ) represents finding P(x) n |Y n The value of the independent variable x is the maximum value. n .

[0038] Step 5 includes: obtaining the brake pedal input push rod displacement through the displacement difference sensor, inputting it into the personalized power-assisted braking characteristic MAP curve to look up the table to obtain the displacement difference between the main and secondary surfaces of the feedback disc, calculating the target position of the power-assisted motor, and inputting it into the iBooster power-assisted motor three-closed-loop control strategy to realize the torque following control of the power-assisted motor.

[0039] In step 5, the brake pedal input push rod displacement x is calculated using the following formula. pedal :

[0040]

[0041] Where, θ m To assist the motor in turning, P h To assist the lead of the motor lead screw and nut pair, i s To assist the transmission ratio of the motor reduction mechanism, Δx sensor Input push rod displacement difference signal to brake pedal; target position x of booster motor * pmsm The calculation formula is:

[0042] x * pmsm =x pedal +Δx,

[0043] Where Δx is the difference in displacement between the primary and secondary surfaces of the feedback disk obtained from the table lookup.

[0044] In step 5, the iBooster motor three-closed-loop control strategy includes: based on the electronic power-assisted braking system, the power-assisted motor responds to the control commands of the electronic control unit (ECU), outputs the power-assisted torque required for braking, and transmits it to the vehicle braking system through the transmission mechanism to generate the corresponding braking force, providing brake pedal characteristics corresponding to the driver's braking personality, and completing the braking action.

[0045] Compared with the prior art, the present invention has the following beneficial effects:

[0046] (1) More refined driver braking personality types, no need for driver to manually select, can automatically identify and match personalized braking assist characteristics, reducing the time and difficulty for car manufacturers to customize electronic power assist braking systems;

[0047] (2) Parametric design of five types of personalized power-assisted braking characteristic MAP curves: safety, comfort, normal, sport, and adventure, to fully meet the driver's personalized brake pedal feel requirements.

[0048] (3) The addition of a Bayesian fusion decision-making mechanism has improved the credibility of current driver braking personality type identification and comprehensively enhanced the level of intelligent human-vehicle interaction. Attached Figure Description

[0049] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments, and the advantages of the present invention in the above and / or other aspects will become clearer.

[0050] Figure 1 This is a schematic diagram of the overall technical solution of the present invention.

[0051] Figure 2 This is a system structure diagram of the iBooster electronic power-assisted braking system used in this invention.

[0052] Figure 3 This is a MAP diagram illustrating the personalized power-assisted braking characteristics described in this invention.

[0053] Figure 4 This is a structural diagram of the BP neural network model described in this invention.

[0054] Figure 5 This is a diagram of the three-closed-loop control strategy for the iBooster motor described in this invention.

[0055] Figure 6 This is a diagram showing the pedal characteristics of the brake-by-wire system described in this invention. Detailed Implementation

[0056] This embodiment provides a method for designing pedal characteristics of a brake-by-wire system based on online recognition of braking personality. It can automatically identify the driver's braking personality type online and adjust the personalized brake assist control strategy according to different driver braking characteristic data, meeting the driving operation needs of different types of drivers, ensuring driving safety, and providing drivers with a better driving experience. The method flow is as follows: Figure 1 As shown in the figure, the electronic power-assisted braking system structure adopted in this invention is as follows: Figure 2 As shown, it includes the following steps:

[0057] S1. Set up a preliminary database to store the full-condition driving braking dataset of the driver driving the target test vehicle and the online identification results of the driver's braking personality type; in the vehicle controller, use deviation standardization to perform dimensionality reduction operation on the driving braking data, and use principal component analysis to extract braking feature data k1~k4 that can characterize the driver's braking personality.

[0058] Specifically, the full-condition driving braking data of the target test vehicle driven by 40 drivers includes: driving braking data collected by radar and camera sensors, including the longitudinal speed of the vehicle in front and the relative distance between the vehicle and the vehicle in front; driving braking data collected by vehicle CAN bus communication, including the longitudinal speed of the vehicle, braking deceleration, brake pedal travel, brake master cylinder pressure and battery status; and the braking characteristic data k1 to k4 extracted to characterize the driver's braking personality, including: k1 vehicle braking deceleration, k2 brake master cylinder pressure, k3 longitudinal relative speed between the vehicle and the vehicle in front, and k4 relative distance between the vehicle and the vehicle in front.

[0059] S2. Extract braking feature data from the full-condition driving braking dataset of the target experimental vehicle driven by the driver to form a training sample set S. Apply the K-Means algorithm to the training sample set S to perform cluster analysis on the driver's braking personality. The clustering results are shown in Table 1, which details five types of driver braking personality. The obtained driver braking personality types include: conservative type I, conservative type II, normal type, aggressive type I, and aggressive type II.

[0060] Table 1

[0061]

[0062]

[0063] Specifically, based on the braking characteristic data analysis of the five cluster centers in the clustering results, the following conclusions were drawn: Conservative Type I drivers are the most cautious and prudent type, prone to tension and alertness during driving, and habitually maintain the greatest distance from the vehicle in front and drive at the slowest speed; Conservative Type II drivers are relatively calm and cautious, prioritizing safety during driving, maintaining a relatively large distance from the vehicle in front and driving at a slower speed, and are more relaxed compared to Conservative Type I; Average drivers are emotionally stable and composed, with driving habits between conservative and aggressive types, and are the most common driver type; Aggressive Type I drivers are relatively agile and energetic, seeking a sense of sportiness and speed during driving, and habitually maintain a close distance from the vehicle in front and drive at a faster speed; Aggressive Type II drivers are the most impatient and easily angered type, seeking adventure and excitement during driving, and habitually maintain the closest distance from the vehicle in front and drive at the fastest speed;

[0064] Specifically, the process of K-Means clustering analysis is as follows:

[0065] (1) Randomly select five cluster centers s from the training sample set S. r =(s1,s2,s3,s4,s5);

[0066] (2) The remaining data samples, excluding the five cluster centers, are assigned to a data cluster formed by one of the five cluster centers;

[0067] (3) Recalculate the cluster centers s of the five data clusters based on the mean principle. r ;

[0068] (4) Calculate the measure function D, where Where m represents the total number of training samples; r represents the r-th personality label, r = 1, 2, 3, 4, 5 corresponding to conservative type I, conservative type II, normal type, radical type I, and radical type II, respectively; s i Let s represent the i-th (i = 1, 2, ..., m) data point in the training samples. i=(k1,k2,k3,k4); s r Indicates the cluster center;

[0069] (5) If the measure function D converges, the cluster analysis ends; if the measure function D does not converge, return to step (2) and continue to execute the above steps (2) to (4) until D converges, and the cluster analysis ends.

[0070] After the cluster analysis, all data samples were assigned a personality label. There are five personality labels: Conservative Type I, Conservative Type II, Normal Type, Radical Type I, and Radical Type II.

[0071] S3. Based on five different driver braking styles, five personalized MAP curves for assisted braking characteristics are designed parametrically, such as... Figure 3 As shown, different brake pedal characteristics are provided, x in the figure. pedal The input push rod displacement is the brake pedal input, and Δx is the displacement difference between the main and secondary surfaces of the feedback disc.

[0072] Specifically, the personalized power-assisted braking characteristic MAP curve represents different types of braking characteristics by setting the following parameters of the electronic power-assisted braking system: linear stage assist ratio, jump increment, and inflection point master cylinder pressure. The linear stage assist ratio is adjusted by changing the displacement difference between the primary and secondary surfaces of the feedback disc during the assist process. Different assist ratios are obtained by setting different displacement differences between the primary and secondary surfaces corresponding to different input push rod displacements. The jump increment is adjusted by regulating the correspondence between the displacement difference between the primary and secondary surfaces of the feedback disc and the input push rod displacement during the jump increment stage. For the same input push rod displacement, the larger the displacement difference between the primary and secondary surfaces of the feedback disc, the larger the corresponding jump increment. The assist characteristic curve of a traditional vacuum booster is selected as the conventional assist characteristic curve, with a linear stage assist ratio of 7.6, a jump increment of 16 bar, and an inflection point master cylinder pressure of 90 bar. Using this curve as the design benchmark, the parameter values ​​are adjusted upwards and downwards to correspond to the safety, comfort, sport, and adventure assist characteristic curves, respectively.

[0073] The aforementioned safety, comfort, normal, sport, and adventure brake assist characteristic curves achieve different brake pedal characteristics and provide different types of brake pedal feel. Specifically,

[0074] The safety-type power assist characteristic curve has the largest jump value and the largest linear stage of assistance, which is suitable for conservative Type I drivers. The braking process is very easy and safe, and the driver can slow down the car to a stop with just a light touch of the pedal.

[0075] The comfort-type power assist curve has a larger jump increase and a larger linear stage of assistance, which is suitable for conservative type II drivers. The braking process is relatively smooth, and the driver does not need to exert much pedal force to decelerate the car to a stop.

[0076] The standard power assist characteristic curve has a boost ratio between the jump and linear stages, which is between comfort and sport, making it suitable for average drivers and a relatively conventional braking mode.

[0077] The sport-type power assist curve has a smaller jump increase and a smaller linear stage assist, making it suitable for aggressive Type I drivers. The braking process is relatively "rough", requiring the driver to press the brake pedal with greater force and depth.

[0078] The adventure-type power assist characteristic curve has the smallest jump increase and the smallest linear stage assist, which is suitable for aggressive Type II drivers. The braking process is very "heavy", and the driver needs to press the brake pedal hard and quickly to trigger the braking.

[0079] The parameters of the electronic power-assisted braking system, based on the MAP curves of different types of personalized power-assisted braking characteristics, are shown in Table 2 below:

[0080] Table 2

[0081]

[0082]

[0083] S4. During the driver's driving process, the driver's braking personality type is identified online using a BP neural network model based on the vehicle's braking data (BP (back propagation) neural network is a concept proposed by scientists led by Rumelhart and McClelland in 1986, and is a multi-layer feedforward neural network trained according to the error back propagation algorithm). Combined with the driver's braking personality type identification results stored in the previous database, the Bayesian fusion decision method is applied to determine the current driver's braking personality type signal with credibility, and it is stored in the previous database.

[0084] Specifically, the structure diagram of the BP neural network model is as follows: Figure 4 As shown. In this embodiment, the number of nodes in the input layer and output layer of the BP neural network model are 4 inputs and 1 output, respectively. The input layer consists of: vehicle braking deceleration, brake master cylinder pressure, longitudinal relative speed between the vehicle and the vehicle in front, and relative distance between the vehicle and the vehicle in front. The output layer is the corresponding driver braking personality type, namely one of conservative type I, conservative type II, normal type, aggressive type I, and aggressive type II.

[0085] The BP neural network model uses the BP algorithm to calculate the errors of the input layer, hidden layer, and output layer, and counts the number of iterations. The maximum number of iterations is set to 1000, the learning rate is 0.01, and the error performance is 0.001.

[0086] Specifically, the driver braking personality type identified online through the BP neural network model is used as the driver braking personality vector group under the current operating condition. This is a single driver braking personality judgment result and cannot fully represent the driver type. The single judgment result is stored in the pre-existing database, and a Bayesian fusion decision method is used for fusion decision-making to finally determine a driver braking personality type with credibility.

[0087] Specifically, the process of applying Bayesian fusion decision-making is as follows:

[0088] The driver braking personality type y identified in the BP neural network during the nth driving process n And the set of driver braking personality types {y1y2y3…y1} stored in the previous database from the first n-1 online identifications. n-1 Synthesize a new data cluster Y n ={y1y2y3…y n-1 y n Then, a Bayesian fusion decision is performed to ultimately determine the driver's braking personality type x for that particular incident. n x n The values ​​from 1 to r correspond to r types of driver braking personality, with r = 5; additionally, r = 0 indicates a type that is difficult to distinguish.

[0089] Based on Bayes' theorem:

[0090] Where P represents the probability of the event; P(x) n |y n ) represents the driver braking personality type identified by the nth BP neural network, y. n Under the premise of fusion decision-making, the driver braking personality type is determined to be x. n The posterior probability; P(x) n |y n ,Y n-1 ) represents the driver braking personality type identified by the nth BP neural network, y. n Under the condition of fusion decision determination, the driver braking personality type is x n And the set of braking personality types of the drivers identified online in the previous n-1 times is Y. n-1 The probability of simultaneous occurrence; P(y) n |x n ,Y n-1 Given a driver braking personality type of x n Under the assumption that y occurs n And the set of braking personality types of the drivers identified online in the previous n-1 times is Y. n-1 The probability of both occurring simultaneously, P(y) n |Y n-1Y represents the set of driver braking personality types identified in the first n-1 online tests. n-1 Under the premise of y n The probability is set as follows, based on the characteristics of the driver braking personality type sample:

[0091] (1) The classification results are mutually independent, i.e., P(y n |x n ,Y n-1 )=P(y n |x n );

[0092] (2) Driver's braking personality type x n It conforms to a Markov process, that is... The classification results are independent of each other, and the driver's braking personality type is x n Substituting the results conforming to the MarKov process into Bayes' theorem, we get: Where P(y) n |x n () indicates that, given a driver braking personality type of x n Observing y under the assumption of class n The probability of P(x). n =i)(i=0,1,…,K) are unknown, and are assumed to be equal probabilities, i.e., P(x n =i) = 1 / (K+1), therefore, Among them, conf(x n The driver braking personality type provided by the classifier x n The confidence level.

[0093] P(x n |x n-1 () indicates the driver's braking personality type x n-1 The driver's braking personality type is x n In an ideal driving environment, the probability of a driver's braking personality type being consistent should be: However, due to differences in road conditions, driver fatigue, distraction, and other factors, the driver's braking personality may deviate from its normal braking personality. Therefore, handling tolerance γ and style tolerance ε are introduced, and the final expression is: `max` represents the maximum value, and `min` represents the minimum value; based on the experiment, 0 ≤ ε ≤ 0.5 and 0 ≤ γ ≤ K; P(x n-1 |Y n-1 The result of the previous fusion decision calculation is given, with the initial condition set as equal chances for all style types. In summary, for x n =0,1,2,…K, calculate the posterior probability P(x)n |Y n ), which yields the output vector [P(0|Y n )P(1|Y n )…P(K|Y n )], corresponding to the maximum P(x n |Y n The category is the current driver braking personality type c. n =argmaxP(x n |Y n ), where argmax(f(x)) is a function that finds the independent variable x that maximizes the value of the function f(x) within the parentheses of argmax(). In this case, it is to find the value of P(x). n |Y n The value of the independent variable x is the maximum value. n .

[0094] S5. The electronic power-assisted braking system ECU receives the current driver's braking personality type signal, matches the personalized power-assisted braking characteristic MAP curve through the personalized braking assist control module, obtains the target position of the assist motor, inputs it into the electronic power-assisted braking system motor three-closed-loop control strategy, and realizes the personalized power-assisted braking function through the vehicle execution module, providing brake pedal characteristics corresponding to the driver's braking personality type.

[0095] Specifically, the brake pedal input push rod displacement can be obtained through a displacement difference sensor. This displacement is then input into the personalized power-assisted braking characteristic MAP curve to look up the displacement difference between the primary and secondary surfaces of the feedback disc. The target position of the power-assisted motor is calculated and input into the iBooster power-assisted motor three-loop control strategy. The iBooster power-assisted motor three-loop control is as follows: Figure 5 As shown, torque following control of the assist motor is achieved. In the figure, x * pmsm N r * I q * These are the target position, target speed, and target quadrature-axis stator current of the booster motor, respectively. pmsm N r U q U d I q I d These are the actual position of the assist motor, actual quadrature-axis voltage, actual direct-axis voltage, actual speed, actual quadrature-axis stator current, and actual direct-axis stator current, respectively.

[0096] Specifically, the brake pedal input push rod displacement x pedal The calculation formula is: Where, θ m To assist the motor in turning, Ph To assist the lead of the motor lead screw and nut pair, i s To assist the transmission ratio of the motor reduction mechanism, Δx sensor Input push rod displacement difference signal to brake pedal; target position x of booster motor * pmsm The calculation formula is: x * pmsm =x pedal +Δx, where Δx is the difference in displacement between the primary and secondary surfaces of the feedback disk obtained from the table;

[0097] Specifically, based on the electronic power-assisted braking system in the vehicle's execution module, the power assist motor responds quickly to the control commands of the ECU, outputting the required braking torque. This torque is transmitted through a transmission mechanism to the vehicle's braking system to generate a corresponding braking force, providing brake pedal characteristics corresponding to the driver's braking personality, and completing the braking action. The pedal characteristic results of the brake-by-wire system described in this invention are as follows: Figure 6 As shown.

[0098] In its specific implementation, this application provides a computer storage medium and a corresponding data processing unit. The computer storage medium is capable of storing a computer program, which, when executed by the data processing unit, can run the invention's content regarding the design method for pedal characteristics of a brake-by-wire system based on online brake personality recognition, as well as some or all of the steps in various embodiments. The storage medium can be a magnetic disk, optical disk, read-only memory (ROM), or random access memory (RAM), etc.

[0099] Those skilled in the art will clearly understand that the technical solutions in the embodiments of the present invention can be implemented using computer programs and their corresponding general-purpose hardware platforms. Based on this understanding, the technical solutions in the embodiments of the present invention, or the parts that contribute to the prior art, can be embodied in the form of computer programs, i.e., software products. These computer program software products can be stored in a storage medium and include several instructions to cause a device containing a data processing unit (which may be a personal computer, server, microcontroller, MUU, or network device, etc.) to execute the methods described in various embodiments or certain parts of the embodiments of the present invention.

[0100] This invention provides a method for designing pedal characteristics of a brake-by-wire system based on online identification of braking characteristics. Many methods and approaches exist for implementing this technical solution; the above description is merely a preferred embodiment of the invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of this invention, and these improvements and modifications should also be considered within the scope of protection of this invention. All components not explicitly stated in this embodiment can be implemented using existing technologies.

Claims

1. A method for designing pedal characteristics of a brake-by-wire system based on online identification of braking characteristics, characterized in that, Includes the following steps: Step 1: Set up a preliminary database to store the full-condition driving braking dataset of the target test vehicle driven by the driver and the online identification results of the driver's braking personality type. In the vehicle controller, deviation standardization is used to perform dimensionality reduction on the driving braking data, and principal component analysis is used to extract braking feature data that can characterize the driver's braking personality, forming a training sample set S. Step 2: Apply the K-Means algorithm to the training sample set S to perform cluster analysis on the driver's braking personality and classify the driver's braking personality type; Step 3: Based on the driver's braking personality type, design a personalized MAP curve for the power-assisted braking characteristic using parametric design to provide different brake pedal characteristics; In step 3, the personalized power-assisted braking characteristic MAP curve represents different types of braking characteristics by setting the following parameters of the electronic power-assisted braking system: linear stage power-up ratio, jump increment, and inflection point master cylinder pressure. The linear stage power-up ratio is adjusted by changing the displacement difference between the primary and secondary surfaces of the feedback disc during the power-up process. Different power-up ratios are obtained by setting different displacement differences between the primary and secondary surfaces corresponding to different input push rod displacements. The jump increment adjustment is achieved by adjusting the correspondence between the displacement difference between the primary and secondary surfaces of the feedback disc and the input push rod displacement during the jump increment stage. For the same input push rod displacement, the larger the displacement difference between the primary and secondary surfaces of the feedback disc, the larger the corresponding jump increment. The power-up characteristic curve of a traditional vacuum booster is selected as the conventional power-up characteristic curve. Step 4: During the driver's driving process, the driver's braking personality type is identified online using a BP neural network model based on the vehicle braking data. Combined with the driver braking personality type identification results stored in the previous database, the Bayesian fusion decision method is used to determine the current driver's braking personality type signal with credibility and store it in the previous database. Step 5: The Electronic Assisted Braking System (ECU) receives the current driver's braking personality type signal, matches the personalized assisted braking characteristic MAP curve, realizes personalized assisted braking function, and provides brake pedal characteristics corresponding to the driver's braking personality type. Step 5 includes: obtaining the brake pedal input push rod displacement through the displacement difference sensor, inputting it into the personalized power-assisted braking characteristic MAP curve to look up the table to obtain the displacement difference between the main and secondary surfaces of the feedback disc, calculating the target position of the power-assisted motor, and inputting it into the iBooster power-assisted motor three-closed-loop control strategy to realize the torque following control of the power-assisted motor; In step 5, the iBooster motor three-closed-loop control strategy includes: based on the electronic power-assisted braking system, the power-assisted motor responds to the control commands of the electronic control unit (ECU), outputs the power-assisted torque required for braking, and transmits it to the vehicle braking system through the transmission mechanism to generate the corresponding braking force, providing brake pedal characteristics corresponding to the driver's braking personality, and completing the braking action.

2. The method according to claim 1, characterized in that, In step 1, the full-condition driving braking data of the driver-driven target test vehicle includes the following data: driving braking data collected by radar and camera sensors, including the longitudinal speed of the vehicle in front and the relative distance between the vehicle and the vehicle in front; driving braking data collected by vehicle CAN bus communication, including the longitudinal speed of the vehicle, braking deceleration, brake pedal travel, brake master cylinder pressure and battery status. The extracted braking feature data that characterizes the driver's braking personality includes: vehicle braking deceleration k1, master cylinder pressure k2, longitudinal relative velocity between the vehicle and the vehicle in front k3, and relative distance between the vehicle and the vehicle in front k4. A training sample set S is constructed using this braking feature data. The training sample set S contains m samples, where m is a natural number. Each sample contains a set of braking feature data characterizing the driver's braking personality. i Let s represent the i-th data point in the training sample S, where i = 1, 2, ..., m. i =(k i1 ,k i2 ,k i3 ,k i4 ), k i1 ,k i2 ,k i3 ,k i4 These represent the vehicle's braking deceleration, master cylinder pressure, longitudinal relative speed between the vehicle and the vehicle in front, and relative distance between the vehicle and the vehicle in front, respectively, in the i-th data in the training sample S.

3. The method according to claim 2, characterized in that, In step 2, the K-Means algorithm is applied to cluster the m samples in the training sample set S into 5 classes, and then the following steps are performed: Step 2-1: Randomly select five cluster centers s from the training sample set S. r r takes values ​​from 1 to 5; Step 2-2: The remaining data samples, excluding the five cluster centers, are assigned to the data clusters formed by one of the five cluster centers. Steps 2-3: Recalculate the cluster centers s of the five data clusters based on the mean principle. r ; Steps 2-4: Calculate the measure function D: Where m represents the total number of training samples; r represents the r-th personality label, and the values ​​of r, 1, 2, 3, 4, and 5 correspond to conservative type I, conservative type II, normal type, radical type I, and radical type II, respectively. Step 2-5: If the measure function D converges, the cluster analysis ends; if the measure function D does not converge, return to step 2-2 and continue executing steps 2-2 to 2-4 until D converges, at which point the cluster analysis ends. After the cluster analysis, all data samples were assigned a personality label. There are five personality labels: Conservative Type I, Conservative Type II, Normal Type, Radical Type I, and Radical Type II.

4. The method according to claim 3, characterized in that, In step 3, five personalized assisted braking characteristic MAP curves are parametrically designed, including safety, comfort, normal, sport, and adventure assisted braking characteristic curves, to achieve different brake pedal characteristics and provide different types of brake pedal feel.

5. The method according to claim 4, characterized in that, In step 3, the safety-type power assist characteristic curve shows the largest jump increase and the largest assist in the linear stage, which is suitable for conservative type I drivers; The comfort-type power steering characteristic curve has a larger jump increase and a larger linear stage of power assistance, making it suitable for conservative Type II drivers. The standard power assist characteristic curve has a boost and linear phase power assist ratio that falls between comfort and sport, making it suitable for average drivers. The sport-type power assist curve has a smaller jump increase and a smaller linear stage of power assist, making it suitable for aggressive Type I drivers. The adventure-type boost curve has the smallest jump increase and the smallest linear boost, making it suitable for aggressive Type II drivers.

6. The method according to claim 5, characterized in that, Step 4 includes: The driver braking personality type y identified in the BP neural network during the nth driving process n And the set of driver braking personality types {y1y2y3…y1} stored in the previous database from the first n-1 online identifications. n-1 Synthesize a new data cluster Y n ={y1y2y3…y n-1 y n Then, a Bayesian fusion decision is performed to ultimately determine the driver's braking personality type x for that particular incident. n x n The values ​​from 1 to r correspond to r types of driver braking personality, with r = 5; additionally, r = 0 indicates a type that is difficult to distinguish. Based on Bayes' theorem: Where P represents the probability of the event; P(x) n |y n ) represents the driver braking personality type identified by the nth BP neural network, y. n Under the premise of fusion decision-making, the driver braking personality type is determined to be x. n The posterior probability; P(x) n |y n ,Y n-1 ) represents the driver braking personality type identified by the nth BP neural network, y. n Under the condition of fusion decision determination, the driver braking personality type is x n And the set of braking personality types of the drivers identified online in the previous n-1 times is Y. n-1 The probability of simultaneous occurrence; P(y) n |x n ,Y n-1 Given a driver braking personality type of x n Under the assumption that y occurs n And the set of braking personality types of the drivers identified online in the previous n-1 times is Y. n-1 The probability of both occurring simultaneously, P(y) n |Y n-1 Y represents the set of driver braking personality types identified in the first n-1 online tests. n-1 Under the premise of y n The probability is set as follows, based on the characteristics of the driver braking personality type sample: The classification results are independent: P(y) n |x n ,Y n-1 )=P(y n |x n ); Driver's braking personality type x n Conforms to a Markov process: The classification results are independent of each other, and the driver's braking personality type is x n Substituting the results conforming to the MarKov process into Bayes' theorem, we get: Where P(y) n |x n () indicates that, given a driver braking personality type of x n Observing y under the assumption of class n The probability of P(x); due to P(x) n =i)(i=0,1,…,K) are unknown, assumed to be equal probabilities, P(x n =i) = 1 / (K+1), therefore, Among them, conf(x n The driver braking personality type provided by the classifier x n Confidence level; P(x n |x n-1 () indicates the driver's braking personality type x n-1 The driver's braking personality type is x n The probability, Introducing manipulation tolerance γ and style tolerance ε, the final expression is: `max` retrieves the maximum value, and `min` retrieves the minimum value. Set 0 ≤ ε ≤ 0.5 and 0 ≤ γ ≤ K; P(x n-1 |Y n-1 The result of the previous fusion decision calculation is given, with the initial condition that all style types have equal chances: For x n =0,1,2,…K, calculate the posterior probability P(x) n |Y n ), thus obtaining the output vector [P(0|Y n )P(1|Y n )…P(K|Y n )], corresponding to the maximum P(x n |Y n The category is the current driver braking personality type c. n =argmaxP(x n |Y n ), where argmaxP(x n |Y n ) represents finding P(x) n |Y n The value of the independent variable x is the maximum value. n .

7. The method according to claim 6, characterized in that, In step 5, the brake pedal input push rod displacement x is calculated using the following formula. pedal : Where, θ m To assist the motor in turning, P h To assist the lead of the motor lead screw and nut pair, i s To assist the transmission ratio of the motor reduction mechanism, Δx sensor Input push rod displacement difference signal to brake pedal; target position x of booster motor * pmsm The calculation formula is: x * pmsm =x pedal +Δx, Where Δx is the difference in displacement between the primary and secondary surfaces of the feedback disk obtained from the table lookup.

Images