Vehicle traction control method

Abstract

The invention relates to the technical field of traction force control systems, and provides a vehicle traction force control method which comprises the steps that vehicle dynamics information of a to-be-controlled vehicle is obtained, and the vehicle dynamics information is information describing the movement condition of the to-be-controlled vehicle; according to the vehicle dynamics information, a plurality of road adhesion coefficients are calculated through a plurality of adhesion coefficient calculation modes; fusing the plurality of pavement adhesion coefficients to obtain a comprehensive pavement adhesion coefficient; according to the vehicle dynamics information, multiple vehicle slip rates are calculated through multiple slip rate calculation modes; fusing the plurality of vehicle slip rates to obtain a comprehensive vehicle slip rate; and according to the comprehensive road adhesion coefficient and the comprehensive vehicle slip rate, the driving torque and the wheel braking force are calculated respectively, and the to-be-controlled vehicle is controlled based on the driving torque and the wheel braking force. According to the method, multiple influence factors are comprehensively considered through data fusion processing, and the control precision of the vehicle traction control system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of traction control systems, and in particular to a vehicle traction control method. Background Art

[0002] Often, on slippery surfaces, such as snow or a pond, the wheels don't generate enough traction to propel the vehicle. Traction Control System (TCS) prevents wheels from spinning due to acceleration. TCS makes better use of the road's adhesion coefficient. When a drive wheel begins to slip, the system helps restore traction by reducing the vehicle's drive torque or applying brake pressure, depending on the degree of slip.

[0003] In the existing technology, the current vehicle-road adhesion coefficient is generally calculated using the vehicle tire force to obtain a reasonable slip rate range. The actual slip rate when the vehicle slips is then calculated, and the vehicle drive torque and wheel braking force are controlled based on a feedback control method to control the slip rate within a reasonable slip rate range. However, the calculation process of this method does not take into account all the influencing factors, resulting in low control accuracy. In addition, the system control parameters are all written into the program through calibration and cannot be updated and adjusted in real time, which lacks adaptability in certain scenarios. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a vehicle traction control method to solve the above technical problems.

[0005] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: A vehicle traction control method, comprising: obtaining vehicle dynamics information of a vehicle to be controlled, the vehicle dynamics information being information describing the movement of the vehicle to be controlled; calculating a plurality of road adhesion coefficients according to the vehicle dynamics information by using a plurality of adhesion coefficient calculation methods; fusing the plurality of road adhesion coefficients to obtain a comprehensive road adhesion coefficient; calculating a plurality of vehicle slip rates according to the vehicle dynamics information by using a plurality of slip rate calculation methods; fusing the plurality of vehicle slip rates to obtain a comprehensive vehicle slip rate; and calculating a driving torque and a wheel braking force respectively according to the comprehensive road adhesion coefficient and the comprehensive vehicle slip rate, so as to control the vehicle to be controlled based on the driving torque and the wheel braking force.

[0006] The present invention has the following beneficial effects: it can calculate the road adhesion coefficient separately through multiple methods, then fuse the individual road adhesion coefficients to obtain a comprehensive road adhesion coefficient. It can also calculate the vehicle slip rate separately through multiple methods, then fuse the individual vehicle slip rates to obtain a comprehensive vehicle slip rate. Finally, the driving torque and vehicle control force are calculated based on the comprehensive road adhesion coefficient and comprehensive vehicle slip rate to control the vehicle. This method, through data fusion, comprehensively considers multiple influencing factors, improves the control accuracy of the vehicle traction control system, and enables adaptive parameter adjustment.

[0007] On the basis of the above technical solution, the present invention can also be improved as follows.

[0008] Furthermore, the vehicle dynamics information includes vehicle driving data, which is data generated during the driving of the vehicle to be controlled and reflects the operating state of the vehicle to be controlled.

[0009] Furthermore, the adhesion coefficient calculation method includes: performing denoising and timestamp alignment processing on the vehicle driving data to obtain new vehicle driving data; performing feature extraction on the new vehicle driving data to obtain a first feature, where the first feature is a feature affected by the road adhesion coefficient; and calculating the road adhesion coefficient based on the first feature using a pre-trained gradient boosting tree model.

[0010] Furthermore, the adhesion coefficient calculation method also includes: calculating the tire force and tire vertical load of the vehicle to be controlled according to the vehicle driving data; and calculating the road adhesion coefficient according to the tire force and the tire vertical load.

[0011] Furthermore, the vehicle dynamics information also includes a road surface image, which is an image of the road surface on which the vehicle to be controlled is traveling; the adhesion coefficient calculation method also includes: performing image preprocessing on the road surface image to obtain a new road surface image; and calculating the road adhesion coefficient based on the new road surface image using a pre-trained image recognition processing model.

[0012] Furthermore, the slip rate calculation method includes: constructing a vehicle dynamics model based on a preset vehicle longitudinal motion equation, a vehicle rotational motion equation and a tire longitudinal force model; and calculating the vehicle slip rate according to the vehicle dynamics model and the vehicle driving data.

[0013] Furthermore, the slip rate calculation method includes: constructing features based on the vehicle driving data to obtain a second feature, where the second feature is a feature affected by the vehicle slip rate; filtering and timestamp aligning the vehicle driving data to obtain processed vehicle driving data; standardizing the second feature to obtain a processed second feature; and calculating the vehicle slip rate using a pre-trained random forest model based on the processed second feature and the processed vehicle driving data.

[0014] Furthermore, the driving torque and wheel braking force are calculated respectively based on the comprehensive road adhesion coefficient and the comprehensive vehicle slip rate, including: determining a target slip rate based on the comprehensive road adhesion coefficient; calculating the difference between the target slip rate and the comprehensive vehicle slip rate to obtain a slip rate control deviation; calculating a first driving torque and a first wheel braking force respectively through PID control and fuzzy logic control based on the slip rate control deviation; calculating a second driving torque and a second wheel braking force respectively through model predictive control and adaptive control; fusing the first driving torque and the second driving torque to obtain the driving torque; and fusing the first wheel braking force and the second wheel braking force to obtain the wheel braking force.

[0015] Furthermore, the first driving torque is calculated in the following manner: calculating the deviation change rate based on the slip rate control deviation; converting the slip rate control deviation and the deviation change rate into fuzzy linguistic variables based on a preset variable fuzzy set and membership function to obtain a slip rate control deviation language value and a deviation change rate language value; determining a PID parameter adjustment language value based on the slip rate control deviation language value and the deviation change rate language value based on a preset fuzzy rule base; defuzzifying the PID parameter adjustment language value by a center of gravity method to obtain a PID parameter adjustment value; calculating a PID output torque based on the PID parameter adjustment value, the slip rate control deviation and a preset PID parameter initial value; and calculating the first driving torque based on the PID output torque.

[0016] Furthermore, the second driving torque is calculated in the following manner: based on a preset vehicle motion equation and tire longitudinal force model, a vehicle state prediction model is constructed through adaptive control; according to the vehicle state prediction model and a preset performance indicator function, a target control sequence is calculated through a solver; and according to the target control sequence, the second driving torque is determined. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 The flowchart of a vehicle traction control method of the present invention is shown. DETAILED DESCRIPTION

[0018] The principles and features of the present invention are described below. The examples given are only used to explain the present invention and are not used to limit the scope of the present invention.

[0019] like Figure 1 As shown, this embodiment provides a vehicle traction control method, including:

[0020] S101. Acquire vehicle dynamics information of a vehicle to be controlled. The vehicle dynamics information is information describing a motion condition of the vehicle to be controlled.

[0021] S102. Calculate multiple road adhesion coefficients using multiple adhesion coefficient calculation methods based on vehicle dynamics information.

[0022] S103: Fusing multiple road adhesion coefficients to obtain a comprehensive road adhesion coefficient.

[0023] S104 : Calculate multiple vehicle slip rates using multiple slip rate calculation methods based on vehicle dynamics information.

[0024] S105 , fusing multiple vehicle slip rates to obtain a comprehensive vehicle slip rate.

[0025] S106 , calculating the driving torque and the wheel braking force respectively according to the comprehensive road adhesion coefficient and the comprehensive vehicle slip rate, so as to control the vehicle to be controlled based on the driving torque and the wheel braking force.

[0026] Optionally, in an embodiment, the vehicle dynamics information includes vehicle driving data, which is data generated during the driving process of the vehicle to be controlled and reflects the operating state of the vehicle to be controlled.

[0027] Optionally, in an embodiment, the adhesion coefficient calculation method further includes: calculating the tire force and tire vertical load of the vehicle to be controlled based on vehicle driving data; and calculating the road adhesion coefficient based on the tire force and tire vertical load.

[0028] This method calculates the road adhesion coefficient through the tire force during vehicle driving. The specific calculation process is as follows:

[0029] The road adhesion coefficient μ is defined as the tire force (longitudinal force F x or lateral force F y ) and tire vertical load F z The ratio of:

[0030]

[0031] Longitudinal adhesion coefficient:

[0032] Lateral adhesion coefficient:

[0033] The required vehicle driving data include: wheel speed ω, lateral and longitudinal acceleration a x and a y , yaw angular velocity Driving torque T, suspension position.

[0034] The calculation of tire force is as follows:

[0035] Longitudinal force F x calculate:

[0036] For driving conditions:

[0037]

[0038] I w is the moment of inertia of the wheel; r is the effective rolling radius of the wheel.

[0039] For braking conditions:

[0040] F x =k b ·P

[0041] Lateral force F y calculate:

[0042]

[0043] m is the vehicle mass; n is the vehicle mass; wheels The number of wheels that are assigned to the lateral force.

[0044] For the vertical load F z The dynamic calculation is as follows:

[0045] The vertical load is affected by vehicle acceleration and suspension characteristics and can be estimated using the following model:

[0046] F z =F z,static +ΔF z,dynamic

[0047] Static load:

[0048] (front axle)

[0049] (rear axle)

[0050] l f 、l r is the distance from the front and rear axles to the center of mass; L is the wheelbase.

[0051] Dynamic load transfer:

[0052] Transfer due to longitudinal acceleration:

[0053] Transfer due to lateral acceleration:

[0054] h is the height of the center of mass; t is the wheelbase.

[0055] Dynamic coupling correction: Tire forces have longitudinal and lateral coupling effects, which require correction in conjunction with the friction ellipse model:

[0056]

[0057] μ max,x and μ max,y are the maximum longitudinal and lateral adhesion coefficients under the current road surface.

[0058] Optionally, in an embodiment, the vehicle dynamics information also includes a road surface image, which is an image of the road surface on which the vehicle to be controlled is traveling; the adhesion coefficient calculation method also includes: performing image preprocessing on the road surface image to obtain a new road surface image; and calculating the road adhesion coefficient based on the new road surface image through a pre-trained image recognition processing model.

[0059] Specifically, road images are captured by the vehicle's onboard camera. During the initial model training phase, the onboard camera captures high-definition images of various road surfaces (such as dry asphalt, wet roads, ice and snow, and gravel), while also recording the actual adhesion coefficient (obtained through experimental measurements).

[0060] Before recognizing road images, they must first be preprocessed. This includes denoising and enhancement (eliminating uneven lighting, shadows, and motion blur, such as using histogram equalization and Gaussian filtering), region-of-interest (ROI) extraction (clipping the road surface to exclude interference such as vehicles and the sky), and standardization (adjusting image size and color space (e.g., converting RGB to HSV) to unify the input format).

[0061] Extract key features related to adhesion coefficient from the image, including:

[0062] Texture features: Gray-level co-occurrence matrix (GLCM), which calculates contrast, correlation, energy, homogeneity, and characterizes road roughness; local binary pattern (LBP), which describes local texture patterns and distinguishes between smooth and rough roads.

[0063] Color features: HSV histogram, which analyzes road surface moisture (for example, wet roads have darker colors) or ice and snow cover (high brightness).

[0064] Geometric features: edge detection (Canny operator), detection of microstructures such as cracks and stones, and indirect reflection of friction potential.

[0065] Deep features: Pre-train CNN (such as ResNet, VGG) and extract high-level semantic features (such as ice, snow, and water recognition) through transfer learning.

[0066] For model construction and training, an end-to-end regression model or classification and mapping methods can be used. For the end-to-end regression model, its input is a road surface image, a CNN-based regression network (such as ResNet-50+ fully connected layer), the loss function uses the mean square error (MSE), and the output is the adhesion coefficient. For the classification + mapping model, the road type (dry asphalt, wet, ice and snow) is identified through the classification model, and according to the classification results, it is mapped to the preset adhesion coefficient range (such as dry asphalt μ = 0.8, ice surface μ = 0.1), using a classification network (such as MobileNetV3) and a rule base. The above two models can also be used in combination to output the adhesion coefficient value and the related confidence level. The main formula principles are as follows:

[0067] Relationship between texture and adhesion coefficient: Roughness R is approximately positively correlated with adhesion coefficient:

[0068] μ∝R=f(GLCM contrast, LBP variance)

[0069] Color is related to humidity: the HSV saturation (S) of wet roads decreases and the brightness (V) decreases:

[0070]

[0071] k is the wetting attenuation coefficient, obtained through experiments.

[0072] Deep Learning Regression Model:

[0073] μ=CNN(I)+∈

[0074] I is the input image and ∈ is the noise term.

[0075] Optionally, in an embodiment, the adhesion coefficient calculation method includes: performing denoising and timestamp alignment processing on the vehicle driving data to obtain new vehicle driving data; performing feature extraction on the new vehicle driving data to obtain a first feature, which is a feature affected by the road adhesion coefficient; and calculating the road adhesion coefficient based on the first feature using a pre-trained gradient boosting tree model.

[0076] When calculating the vehicle-road adhesion coefficient based on the vehicle's motion state, the required vehicle driving data include: wheel speed, lateral and longitudinal acceleration, driving and braking torque, vehicle mass, tire radius, center of mass height, and wheelbase, etc.

[0077] First, the data needs to be preprocessed. Specifically, this involves removing high-frequency noise from the wheel speed and acceleration data, aligning the data timestamps to ensure sensor data synchronization, and generating labels to obtain the true adhesion coefficient value (either through experimental calibration or inverse calculation of the dynamic equation).

[0078] Physical and statistical features related to the adhesion coefficient are extracted from the data. The core features include: slip rate, wheel speed difference, torque-acceleration, load transfer characteristics, energy characteristics, etc.

[0079] For the construction of the gradient boosting tree model (XGBoost), specifically:

[0080] Inputs include: slip ratio, wheel speed difference, torque mean, and acceleration variance.

[0081] The output includes: road adhesion coefficient or road type classification; when the output is road type classification, the corresponding road adhesion coefficient needs to be determined based on the road type classification.

[0082] Model configuration: objective function (for regression tasks, use mean squared error (MSE) or Huber loss (anti-noise)).

[0083] Training process: The training set, validation set, and test set are divided into layers according to the road type, and the proportion is determined according to the actual situation.

[0084] In this embodiment, the road slip ratio is calculated respectively by the above three methods, and finally the three road adhesion coefficients obtained are fused. Specifically:

[0085]

[0086] Among them, μ Ave is the comprehensive road adhesion coefficient, μ1, μ2, and μ3 are the road adhesion coefficients calculated using the above three methods.

[0087] The calculation using the average value method is relatively simple and fast. For use cases with higher accuracy requirements, the confidence or weight coefficients of the three road adhesion coefficients can be set according to needs, and then the comprehensive road adhesion coefficient can be obtained based on the confidence or weight coefficients of the three road adhesion coefficients.

[0088] The slip rate calculation method includes: calculating the vehicle slip rate based on wheel speed and acceleration. The calculation process is as follows:

[0089] Obtain the angular velocity ω (unit: rad / s) of the vehicle wheel and the effective rolling radius r (unit: m) of the wheel.

[0090] Calculate the wheel linear velocity, wheel linear velocity v wheel =ω×r (unit: m / s).

[0091] Get the vehicle's longitudinal acceleration a (unit: m / s 2 ).

[0092] Integrate the acceleration to find the vehicle speed v. If the initial speed is v0, calculate the actual speed of the vehicle by integrating the acceleration:

[0093]

[0094] Determine the working conditions and select the formula to calculate the slip rate:

[0095] Driving conditions (acceleration, v wheel >v):

[0096]

[0097] Braking condition (deceleration, v>v wheel ):

[0098]

[0099] Optionally, in an embodiment, the slip rate calculation method includes: constructing a vehicle dynamics model based on a preset vehicle longitudinal motion equation, a vehicle rotational motion equation and a tire longitudinal force model; and calculating the vehicle slip rate according to the vehicle dynamics model and vehicle driving data.

[0100] Vehicle longitudinal motion equation:

[0101]

[0102] m vehicle mass; Vehicle longitudinal acceleration; F x Tire longitudinal force; F res Rolling resistance, slope resistance, etc.

[0103] The vehicle's rotational motion equation is:

[0104]

[0105] I w Wheel moment of inertia; Wheel angular acceleration; T driving braking torque.

[0106] Tire longitudinal force model: Tire longitudinal force F x It is a function of the slip rate S, and mainly includes linear models and nonlinear models.

[0107] Linear model: F x =C s S,C s Tire longitudinal stiffness.

[0108] Nonlinear model: F x =D·sin(C·arctan(BS-E(BS-arctan(BS)))), B, C, D, and E are all experimental fitting parameters, which depend on the tire and road surface.

[0109] Vehicle speed estimation:

[0110] Once the vehicle speed v and wheel speed ω are known, substituting them into the above slip rate formula can give the slip rate.

[0111] Optionally, in an embodiment, the slip rate calculation method includes: constructing features based on vehicle driving data to obtain a second feature, where the second feature is a feature affected by the vehicle slip rate; filtering and timestamp aligning the vehicle driving data to obtain processed vehicle driving data; standardizing the second feature to obtain a processed second feature; and calculating the vehicle slip rate through a pre-trained random forest model based on the processed second feature and the processed vehicle driving data.

[0112] The vehicle driving data required for this method includes wheel speed, lateral and longitudinal acceleration, torque, timestamp (for time series feature extraction), etc. Label generation is based on the physical formula of slip rate of the reference dynamic model mentioned above, including driving and braking conditions.

[0113] Key feature structures include:

[0114] Basic characteristics: wheel linear speed (v wheel =ω×r), vehicle acceleration integral speed (v acc =∫a x dt), torque-wheel speed relationship (T / v wheel Responsive power status).

[0115] Time window statistical features: 1) wheel speed mean, variance, and rate of change; 2) acceleration mean, maximum, and minimum; 3) torque integral.

[0116] Dynamic interaction characteristics: 1) Wheel speed difference (the difference between driven and non-driven wheels); 2) Preliminary estimate of slip ratio (initial guess based on physical formulas); 3) Product of acceleration and torque (reflecting instantaneous power).

[0117] Data preprocessing includes: 1) low-pass filtering of acceleration and wheel speed; 2) aligning the sampling timestamps of different sensors; and 3) standardizing the features to eliminate dimension effects.

[0118] Random forest model regression task formula:

[0119]

[0120] Ti (X) is the predicted value of the i-th decision tree; N trees is the number of trees; is the slip rate prediction value.

[0121] Model validation evaluation indicators include mean square error and the coefficient of determination R 2 (Reflects the proportion of variance explained by the model).

[0122] In this embodiment, the vehicle slip ratio is calculated respectively by the above three methods, and finally the three obtained vehicle slip ratios are fused, specifically as follows:

[0123]

[0124] Among them, S Est is the comprehensive vehicle slip rate, and S1, S2, and S3 are the vehicle slip rates calculated by the above three methods respectively.

[0125] Optionally, in an embodiment, the driving torque and the wheel braking force are calculated respectively according to the comprehensive road adhesion coefficient and the comprehensive vehicle slip rate, including: determining the target slip rate according to the comprehensive road adhesion coefficient, which can be specifically set according to engineering experience or demand; calculating the difference between the target slip rate and the comprehensive vehicle slip rate to obtain the slip rate control deviation; calculating the first driving torque and the first wheel braking force respectively through PID control and fuzzy logic control according to the slip rate control deviation; calculating the second driving torque and the second wheel braking force respectively through model predictive control and adaptive control; fusing the first driving torque and the second driving torque to obtain the driving torque; fusing the first wheel braking force and the second wheel braking force to obtain the wheel braking force.

[0126] Optionally, in an embodiment, the first driving torque is calculated in the following manner: calculating the deviation change rate based on the slip control deviation; converting the slip control deviation and the deviation change rate into fuzzy linguistic variables based on a preset variable fuzzy set and membership function to obtain a slip control deviation language value and a deviation change rate language value; determining a PID parameter adjustment language value based on the slip control deviation language value and the deviation change rate language value based on a preset fuzzy rule base; defuzzifying the PID parameter adjustment language value by a center of gravity method to obtain a PID parameter adjustment value; calculating the PID output torque based on the PID parameter adjustment value, the slip control deviation and the preset PID parameter initial value; and calculating the first driving torque based on the PID output torque.

[0127] The main idea of ​​this method is: PID control generates the basic control quantity according to the slip rate deviation, and then the fuzzy logic control dynamically adjusts the PID parameters (K p , Ki , K d ) to cope with certain nonlinear and uncertain vehicle dynamics.

[0128] The calculation formula of slip ratio deviation e is:

[0129] e=S target -S actual

[0130] S target is the target slip ratio, S actual is the vehicle slip rate;

[0131] Deviation change rate The calculation formula is:

[0132]

[0133] The main process of fuzzy logic control is as follows:

[0134] Fuzzification: The input deviation e and the deviation change rate Convert to fuzzy linguistic variables and define membership function.

[0135] Input variable fuzzy set:

[0136] e: {negative big (NB), negative small (NS), zero (ZO), positive small (PS), positive big (PB)};

[0137] {Negative big (NB), negative small (NS), zero (ZO), positive small (PS), positive big (PB)};

[0138] Membership function: triangular or Gaussian function is commonly used.

[0139] Fuzzy rule base: According to relevant requirements, engineering experience or test data, a fuzzy rule table is formulated to define different e and How to adjust PID parameters.

[0140] For example:

[0141] If e is PB and is NB, then ΔKp is PB, ΔKi is NB, and ΔKd is PS.

[0142] Total number of rules = number of input fuzzy sets 2 .

[0143] Fuzzy reasoning and defuzzification:

[0144] The reasoning method adopts Mamdani or Sugeno method to calculate the activation strength of each rule and synthesize the output fuzzy set.

[0145] Defuzzification: Convert fuzzy outputs into precise values, often using the Center of Atmosphere (COA) method:

[0146]

[0147] where μ i is the rule activation strength, K p,i is the corresponding fuzzy set center value.

[0148] Adaptive PID control:

[0149] The basic PID output torque is:

[0150]

[0151] Dynamically adjust parameters through fuzzy logic:

[0152]

[0153] K p0 , K i0 , K d0 are the initial PID parameters, which are determined by calibration.

[0154] Target torque calculation

[0155] T target =T PID +T feedforward

[0156] T feedforward It is feed-forward torque that can be predicted based on the vehicle model and used as needed.

[0157] In order to prevent the integral term from accumulating errors, an anti-saturation mechanism needs to be added, such as integral separation or clamping:

[0158]

[0159] ∈ is the slip rate error threshold, which is set according to requirements.

[0160] The calculation method for the first wheel braking force is the same as the first driving torque calculation method described above: PID control generates a basic control variable based on the slip ratio deviation. Then, fuzzy logic control dynamically adjusts the PID parameters based on the deviation and the rate of change of the deviation. The process for calculating the first wheel braking force is the same as that for calculating the first driving torque, differing only in the specific parameters involved (e.g., Kp, Ki, Kd) and initial values ​​(initial driving torque value, initial wheel braking force value).

[0161] Optionally, in an embodiment, the second driving torque is calculated in the following manner: based on a preset vehicle motion equation and tire longitudinal force model, a vehicle state prediction model is constructed through adaptive control; according to the vehicle state prediction model and a preset performance indicator function, a target control sequence is calculated through a solver; and according to the target control sequence, the second driving torque is determined.

[0162] The main idea of ​​this method is to use model predictive control (MPC) to predict future states based on the vehicle dynamics model and generate the optimal control sequence (target torque) through rolling optimization. Adaptive control is used to estimate system parameters (such as tire-road friction coefficient and vehicle mass) online, and the MPC prediction model is dynamically modified to improve its robustness. The main steps are as follows:

[0163] System modeling and state definition:

[0164] Vehicle dynamics model, longitudinal dynamics ignores suspension and lateral motion, state equation:

[0165]

[0166] Slip definition:

[0167] Braking conditions:

[0168] Driving conditions:

[0169] v vehicle longitudinal velocity; ω wheel angular velocity; F x Tire longitudinal force, F x =μ(λ)F z , μ(λ) tire-road friction coefficient (nonlinear function, requires online evaluation), F z Vertical load; T target moment; m vehicle mass; I w The rotating mass of the wheel; r is the wheel radius.

[0170] The following parameters are uncertain and change with time: road friction characteristics μ(λ), vertical load F z (dynamic load transfer), vehicle mass m. The goal of adaptation is to estimate parameters online and modify the prediction model.

[0171] Adaptive parameter estimation:

[0172] If μ(λ) can be expressed as a linear combination of known basis functions:

[0173] μ(λ)=θ T φ(λ)

[0174] θ is the parameter vector to be estimated (such as polynomial coefficients); φ(λ) is the basis function vector (φ=|λ,λ2 |).

[0175] Based on Lyapunov stability theory, the parameter update law is designed:

[0176]

[0177] Discretized form:

[0178] θ k+ =θ k -γφ(λ k )e k Δt

[0179] e=λ target -λ slip rate tracking error; γ adaptive gain (positive definite matrix, controls the convergence speed).

[0180] Model Predictive Control Design:

[0181] Discrete prediction model, discretize the continuous model (Euler method, step size Δt):

[0182]

[0183] Tire force model (after adaptive correction):

[0184]

[0185] Rolling optimization problem: At each time t, find the optimal control sequence {T0, T1…T N-1}:

[0186]

[0187] Constraints:

[0188] Dynamic constraint: v k+1 =f(v k ,ω k ,T k ,θ k )

[0189] Control input limit: T min ≤T k ≤T max

[0190] Slip rate safety range: 0≤λ k ≤λ max

[0191] Real-time solution implementation: The solver chooses to use sequential quadratic programming (SQP) or interior point method (IPOPT), and only executes the first step control quantity (T target =T0), scroll to the next moment and re-optimize.

[0192] The calculation method for the second wheel braking force is the same as the second driving torque calculation described above: Model predictive control (MPC) uses the vehicle dynamics model to predict future states, generating the optimal control sequence (target wheel control force) through rolling optimization. Adaptive control then estimates system parameters online, dynamically revising the MPC prediction model. The process for calculating the second wheel braking force is identical to that for calculating the second driving torque, differing only in the specific parameters involved.

[0193] The calculation of driving torque and wheel braking force is achieved through the following formula:

[0194] T out =Factor·T1+(1-Factor)·T2

[0195] Among them, T out is the final driving torque (wheel braking force); T1 is the first driving torque (first wheel braking force); T2 is the second driving torque (second wheel braking force), and Factor is the weight ratio of the first driving torque (first wheel braking force). The weight can be set according to subjective engineering experience or obtained by looking up the table based on vehicle dynamic data.

[0196] In summary, this method is particularly suitable for vehicles equipped with front cameras and IBCU (Integrated Braking Control Unit) products. This method can calculate the road adhesion coefficient separately through various methods, and then fuse the individual road adhesion coefficients to obtain a comprehensive road adhesion coefficient. It also calculates the vehicle slip rate separately through various methods, and then fuses the individual vehicle slip rates to obtain a comprehensive vehicle slip rate. Finally, the driving torque and vehicle control force are calculated based on the comprehensive road adhesion coefficient and the comprehensive vehicle slip rate to control the vehicle to be controlled. This method achieves comprehensive consideration of multiple influencing factors through data fusion processing, improves the control accuracy of the vehicle traction control system, and realizes adaptive parameter adjustment.

[0197] The above description is merely a preferred embodiment of the present invention and an illustration of the technical principles employed. Those skilled in the art should understand that the scope of disclosure involved in the present invention is not limited to the technical solutions formed by the specific combination of the above-mentioned technical features, but also includes other technical solutions formed by any combination of the above-mentioned technical features or their equivalents without departing from the above-mentioned disclosed concepts. For example, a technical solution formed by replacing the above-mentioned features with (but not limited to) technical features with similar functions disclosed in the present invention.

[0198] It should be noted that the terms "first," "second," and the like in the specification and claims of this application are used to distinguish similar objects and to define a specific order or precedence. Where appropriate, the order used for similar objects may be interchanged, such that the embodiments of the present application described herein can be implemented in an order other than the order shown or described.

[0199] Those skilled in the art will appreciate that the present invention may be implemented as a system, method, or computer program product. Therefore, the present invention may be implemented in the following forms: entirely in hardware, entirely in software (including firmware, resident software, microcode, etc.), or in a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, the present invention may be implemented in the form of a computer program product embodied in one or more computer-readable media containing computer-readable program code.

[0200] Although the embodiments of the present invention have been shown and described above, it will be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. A person skilled in the art may change, modify, replace and modify the above embodiments within the scope of the present invention.

Claims

1. A vehicle traction control method, characterized in that: include: Acquiring vehicle dynamics information of a vehicle to be controlled, wherein the vehicle dynamics information is information describing a motion condition of the vehicle to be controlled; Calculating a plurality of road adhesion coefficients using a plurality of adhesion coefficient calculation methods according to the vehicle dynamics information; fusing the multiple road adhesion coefficients to obtain a comprehensive road adhesion coefficient; Calculating a plurality of vehicle slip rates using a plurality of slip rate calculation methods according to the vehicle dynamics information; fusing the multiple vehicle slip rates to obtain a comprehensive vehicle slip rate; According to the comprehensive road adhesion coefficient and the comprehensive vehicle slip ratio, a driving torque and a wheel braking force are calculated respectively, so as to control the vehicle to be controlled based on the driving torque and the wheel braking force.

2. The vehicle traction control method according to claim 1, characterized in that: The vehicle dynamics information includes vehicle driving data, which is data generated during the driving of the vehicle to be controlled and reflects the operating state of the vehicle to be controlled.

3. The vehicle traction control method according to claim 2, characterized in that: The adhesion coefficient calculation method includes: Performing denoising and timestamp alignment processing on the vehicle driving data to obtain new vehicle driving data; Performing feature extraction on the new vehicle driving data to obtain a first feature, where the first feature is a feature affected by a road adhesion coefficient; According to the first feature, the road adhesion coefficient is calculated using a pre-trained gradient boosting tree model.

4. The vehicle traction control method according to claim 2, characterized in that: The adhesion coefficient calculation method also includes: Calculating the tire force and tire vertical load of the vehicle to be controlled according to the vehicle driving data; The road adhesion coefficient is calculated based on the tire force and the tire vertical load.

5. The vehicle traction control method according to claim 1, characterized in that: The vehicle dynamics information further includes a road surface image, wherein the road surface image is an image of the road surface on which the vehicle to be controlled is traveling; The adhesion coefficient calculation method also includes: performing image preprocessing on the road surface image to obtain a new road surface image; The road adhesion coefficient is calculated based on the new road surface image using a pre-trained image recognition processing model.

6. The vehicle traction control method according to claim 2, characterized in that: The slip ratio calculation method includes: Construct a vehicle dynamics model based on the preset vehicle longitudinal motion equation, vehicle rotational motion equation, and tire longitudinal force model; The vehicle slip ratio is calculated according to the vehicle dynamics model and the vehicle driving data.

7. The vehicle traction control method according to claim 2, characterized in that: The slip ratio calculation method includes: Performing feature construction based on the vehicle driving data to obtain a second feature, where the second feature is a feature affected by the vehicle slip rate; Performing filtering processing and time stamp alignment processing on the vehicle driving data to obtain processed vehicle driving data; performing a standardization process on the second feature to obtain a processed second feature; The vehicle slip rate is calculated using a pre-trained random forest model according to the processed second feature and the processed vehicle driving data.

8. The vehicle traction control method according to claim 1, characterized in that: The calculating of the driving torque and the wheel braking force according to the comprehensive road adhesion coefficient and the comprehensive vehicle slip rate includes: determining a target slip ratio according to the comprehensive road surface adhesion coefficient; calculating a difference between the target slip ratio and the comprehensive vehicle slip ratio to obtain a slip ratio control deviation; Calculating a first driving torque and a first wheel braking force respectively through PID control and fuzzy logic control according to the slip ratio control deviation; respectively calculating a second driving torque and a second wheel braking force through model predictive control and adaptive control; fusing the first driving torque and the second driving torque to obtain the driving torque; The first wheel braking force and the second wheel braking force are combined to obtain the wheel braking force.

9. The vehicle traction control method according to claim 8, characterized in that: The first driving torque is calculated as follows: Calculating a deviation change rate based on the slip ratio control deviation; Based on a preset variable fuzzy set and membership function, the slip rate control deviation and the deviation change rate are converted into fuzzy linguistic variables to obtain a slip rate control deviation linguistic value and a deviation change rate linguistic value; Based on a preset fuzzy rule library, determining a PID parameter adjustment language value according to the slip rate control deviation language value and the deviation change rate language value; Defuzzification is performed on the PID parameter adjustment language value by using a center of gravity method to obtain a PID parameter adjustment value; Calculating a PID output torque according to the PID parameter adjustment value, the slip rate control deviation, and a preset PID parameter initial value; The first driving torque is calculated according to the PID output torque.

10. The vehicle traction control method according to claim 8, characterized in that: The second driving torque is calculated as follows: Based on the preset vehicle motion equations and tire longitudinal force model, a vehicle state prediction model is constructed through adaptive control; Calculating a target control sequence through a solver based on the vehicle state prediction model and a preset performance indicator function; The second driving torque is determined according to the target control sequence.

Images