Intelligent automobile stability control system based on tire cornering stiffness estimation

Abstract

The invention discloses an intelligent automobile stability control system based on tire cornering stiffness estimation. The intelligent automobile stability control system comprises a driver-vehicle-road comprehensive information acquisition module, a control variable expected value calculation module, a target stabilizing moment calculation module and a moment distribution mode selection module. According to the invention, information acquired by a driver-vehicle-road comprehensive information acquisition module is utilized to calculate a tire cornering stiffness mutation factor and selectively start a tire cornering stiffness dynamic estimator based on a depth deterministic strategy gradient algorithm, so that a control variable expected value is accurately calculated; and calculating a target stable moment by using an adaptive fuzzy integral sliding mode algorithm, and determining a moment distribution mode according to the numerical values of the three factors and a moment distribution mode selection rule. According to the method, the tire cornering stiffness can be estimated in time according to the sudden change degree of the tire cornering stiffness, and the transverse stability control effect of the electric automobile is guaranteed.

Description

Technical Field

[0001] The invention belongs to the field of automobile safety, relates to an active safety control technology for electric vehicles, and specifically relates to an intelligent automobile stability control system based on tire cornering stiffness estimation. Background Art

[0002] As an important way to ensure the active safety of the vehicle, lateral stability control stabilizes the lateral movement of the vehicle by applying driving and braking forces to the inner and outer wheels to generate additional yaw torque. Each motor of the four-wheel hub motor-driven electric vehicle is independently controllable and can quickly switch between driving and braking states. Based on this, continuous and efficient control of the vehicle's lateral movement can be achieved, further improving the driving stability control effect.

[0003] Tire cornering stiffness refers to the ratio between the tire lateral force and the tire slip angle. It characterizes the rigidity of the tire under the action of turning or lateral force, and directly affects the handling and stability of the car. Tire cornering stiffness is affected by factors such as tire load, tire pressure and road adhesion, and is prone to change. However, the existing lateral stability control system uses a linear two-degree-of-freedom car model to obtain the expected value of the control variable under the assumption that the tire cornering stiffness remains unchanged. Under extreme operating conditions, a sudden change in the tire cornering stiffness will cause a deviation in the expected value of the desired control variable, which does not meet the requirements of the vehicle's lateral stability control, thereby reducing the vehicle's lateral stability control effect. Therefore, it is necessary to develop an intelligent vehicle stability control system based on tire cornering stiffness estimation to further improve the active safety of the vehicle. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent vehicle stability control system based on tire cornering stiffness estimation to solve the problems faced in the above-mentioned background technology.

[0005] In order to achieve the above object, the present invention provides the following scheme:

[0006] An intelligent vehicle stability control system based on tire cornering stiffness estimation includes a driver-vehicle-road comprehensive information acquisition module, a control variable expected value calculation module, a target stability torque calculation module and a torque distribution mode selection module;

[0007] The driver-vehicle-road comprehensive information acquisition module is used to acquire the driver operation information, vehicle component use and driving status information and road condition information required to maintain the lateral stability of the vehicle; the driver operation information includes the steering wheel angle and the driver's grip on the steering wheel; the vehicle component use and driving status information includes tire aspect ratio, tire pressure, tire temperature, tire wear depth, tire lateral stiffness factory calibration value, battery charge state, motor power factor, longitudinal speed, vehicle mass, yaw angular velocity, center of mass sideslip angle and front wheel angle; the road condition information includes road adhesion coefficient, road curvature and road roughness;

[0008] The control variable expected value calculation module calculates the expected yaw angular velocity ω using the nonlinear two-degree-of-freedom vehicle model motion differential equation r_d and the desired sideslip angle β d The nonlinear two-degree-of-freedom vehicle model motion differential equation is based on the tire cornering stiffness mutation factor K tcs Establish: The tire cornering stiffness mutation factor K tcs , used to determine whether to start the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm; the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is designed with input state variables s and output action variables a as follows:

[0009]

[0010] In the formula, v x is the longitudinal speed; δ f is the front wheel turning angle; F z is the vertical load on the tire at the contact point; α is the tire slip angle; R tire is the coefficient reflecting the basic characteristics and usage status of the tire; μ is the road adhesion coefficient; C est is the estimated value of tire cornering stiffness;

[0011] The tire basic characteristics and usage status reflection coefficient R tire The calculation formula is:

[0012]

[0013] Wherein, P is the current tire pressure; P0 is the normal tire pressure, which is 2.2 bar; T is the current tire temperature; T0 is the normal tire temperature, which is 40°C; D is the current tire wear depth; D0 is the normal tire wear depth, which is 1.6 mm; τ is the characterization coefficient of the current tire structure type, if the current tire is a common bias tire, the value is 1, if the current tire is a radial tire, the value is 2; τ0 is the characterization coefficient of the common bias tire structure type, the value is 1; F is the current tire aspect ratio; F0 is the reference tire aspect ratio, which is 60%; κ1, κ2, κ3, κ4 and κ5 are weight coefficients, and the specific values ​​are determined by simulation and experiment;

[0014] The tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is designed with a reward function R as follows:

[0015]

[0016] Where C0 is the factory calibration value of the tire cornering stiffness; σ(s) is the standard deviation of the input state variable s; s i is the i-th component of the input state variable s; μ i is the mean value of the i-th component of the input state variable s; σ i is the standard deviation of the i-th component of the input state variable s; p is the penalty term power, which is 2; λ1 and λ2 are weight coefficients, whose specific values ​​are determined through simulation and experiments;

[0017] The target stability torque calculation module is used to calculate the additional yaw torque required to maintain the lateral stability of the vehicle; the additional yaw torque is obtained based on an adaptive fuzzy integral sliding mode algorithm; the adaptive fuzzy integral sliding mode algorithm, wherein the input of the fuzzy algorithm is the vehicle operation feedback coefficient Q vof and the driver operation feedback coefficient Q dof , the output is the integral term coefficient c of the integral sliding surface;

[0018] The torque distribution mode selection module is based on the driving motor state factor S m , power battery state factor S b and the lateral instability characterization factor S us , and combined with the torque distribution method selection rules, determine the torque distribution mode; the torque distribution mode is specifically divided into five modes: distributed motor drive torque distribution mode, distributed motor drive-braking torque distribution mode, distributed motor drive-braking and hydraulic braking combined torque distribution mode, distributed motor drive and hydraulic braking combined torque distribution mode and hydraulic differential braking torque distribution mode.

[0019] The tire cornering stiffness mutation factor K tcsThe branching factor F is used by the vehicle and components. vcu , vehicle driving state branch factor F vrs and vehicle driving condition branching factor F voc constitute;

[0020] The vehicle and components use branching factor F vcu The calculation formula is:

[0021]

[0022] In the formula, t v is the vehicle usage time; t t How long the tires last; w T and w P They are the vehicle usage time weight coefficient, tire usage time weight coefficient, current tire temperature weight coefficient and current tire pressure weight coefficient, and their specific values ​​are determined through simulation and experiments;

[0023] The vehicle driving state branch factor F vrs The calculation formula is:

[0024]

[0025] Where θ is the steering wheel angle; w v 、w θ and w α They are the vertical load weight coefficient at the tire contact point, the longitudinal vehicle speed weight coefficient, the steering wheel angle weight coefficient and the tire slip angle weight coefficient, and the specific values ​​are determined through simulation and experiments;

[0026] The vehicle driving condition branching factor F voc The calculation formula is:

[0027] F voc =w μ ·μ 2 +w ψ ·(ψ+u 2 )+w u ·log(1+u)

[0028] Where μ is the road adhesion coefficient; ψ is the road curvature; u is the road roughness; w μ 、w ψ and w u They are the road adhesion coefficient weight coefficient, road curvature weight coefficient and road roughness weight coefficient, and their specific values ​​are determined through simulation and experiments;

[0029] The tire cornering stiffness mutation factor K tcsThe calculation formula is:

[0030]

[0031] In the formula, w vcu 、w vrs and w voc They are the branch factor weight coefficients for vehicle and component use, vehicle driving state branch factor weight coefficients, and vehicle driving condition branch factor weight coefficients, and their specific values ​​are determined through simulation and testing;

[0032] The tire cornering stiffness mutation factor K tcs , used to determine whether to start the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm. The specific judgment rule is: when the tire cornering stiffness mutation factor K tcs When the tire cornering stiffness mutation factor K0 is greater than or equal to the tire cornering stiffness mutation threshold K0, the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is turned on to obtain the tire cornering stiffness that changes under the current working condition to improve the lateral stability control effect; when the tire cornering stiffness mutation factor K tcs When it is less than the tire cornering stiffness mutation threshold K0, the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is turned off, and the factory-calibrated tire cornering stiffness is used to reduce the workload of the vehicle central controller;

[0033] The specific expression formula of the nonlinear two-degree-of-freedom vehicle model motion differential equation is:

[0034]

[0035] In the formula, C f is the sum of the tire cornering stiffness of the two front wheels; C r is the sum of the tire cornering stiffness of the two rear wheels; ω r is the yaw rate; β is the sideslip angle of the center of mass; a and b are the distances from the center of mass to the front and rear axes respectively; I z is the moment of inertia of the vehicle around the z-axis; and are factory calibration values ​​of the tire cornering stiffness of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and The estimated values ​​of the tire cornering stiffness of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and The tire cornering stiffness mutation factors of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and are the tire cornering stiffness mutation thresholds of the left front wheel, right front wheel, left rear wheel and right rear wheel, all of which are 0.5;

[0036] The desired yaw rate ω r_d and the desired center of mass sideslip angle β d The calculation formula is:

[0037]

[0038] In the formula, is the vehicle stability factor.

[0039] The vehicle operation feedback coefficient Q vof The calculation formula is:

[0040]

[0041] Where D f1 , D fr , D rl and D rr The current tire wear depths of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; x_fl 、F x_fr 、F x_rl and F x_rr are the longitudinal forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; F y_fl 、F y_fr 、F yr_l and F y_rr are the lateral forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; F z_fl 、F z_fr 、F z_rl 、F z_rr are the vertical forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; μ fl , μ fr , μ r1 and μ rr are the road adhesion coefficients of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; k tire and k v is the weight coefficient, and its specific value is determined through simulation and experiments;

[0042] The driver operation feedback coefficient Q dof The calculation formula is:

[0043]

[0044] In the formula, F driver The driver's grip on the steering wheel; is the steering wheel angle change rate; w θ , and are the weight coefficients of steering wheel angle, steering wheel angle change rate and driver’s steering wheel grip force, and their specific values ​​are determined through simulation and experiment;

[0045] The specific design of the sliding surface of the adaptive fuzzy integral sliding mode algorithm is as follows:

[0046] s=ω r -ω r_d +(β-β d )+c∫[ω r -ω r_d +ρ(β-β d )]dt

[0047] Where c is the integral term coefficient of the integral sliding surface, and its specific value is determined based on the vehicle operation feedback coefficient Q vof and the driver operation feedback coefficient Q dof It is obtained through fuzzy reasoning; ρ is the weight coefficient of the control variable, and the specific value is determined through simulation and experiment;

[0048] Using the adaptive fuzzy integral sliding mode algorithm, the additional yaw moment is obtained as:

[0049]

[0050] Where ε and k are sliding reaching law coefficients, and their specific values ​​are obtained through multiple simulations; sat(s) is a saturation function used to weaken chattering.

[0051] The driving motor state factor S m The calculation formula is:

[0052]

[0053] Where PF is the motor power factor; PF max and PF min They are the maximum and minimum values ​​of the motor power factor when the motor is working normally, which are determined through motor testing; T m is the motor temperature; and They are the maximum and minimum values ​​of the motor temperature when the motor is working normally, which are determined through motor tests; H(x) is the Heaviside step function;

[0054] The driving motor state factor S b The calculation formula is:

[0055]

[0056] Where SOC is the battery state of charge; Tb is the battery temperature; and They are the maximum and minimum values ​​of the battery temperature when the battery is working normally, which are determined through battery testing;

[0057] The lateral instability degree characterization factor S us The calculation formula is:

[0058]

[0059] In the formula, a y is the lateral acceleration; β max and They are the maximum values ​​of longitudinal speed, yaw angular velocity, sideslip angle at the center of mass and lateral acceleration when the vehicle is in laterally stable state, which are determined through multiple simulations; ε1, ε2, ε3 and ε4 are weight coefficients, and their specific values ​​are determined through simulation and experiments.

[0060] The torque distribution mode selection rule is specifically as follows:

[0061] When the driving motor state factor S m is equal to 1, and the lateral instability degree characterization factor S us When the lateral instability level is less than a first threshold, the distributed motor drive torque distribution mode is selected; the first threshold of the lateral instability level is determined by multiple simulations;

[0062] When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 1, and the lateral instability characterization factor S us When the lateral instability level is within the range of the first threshold value and the second threshold value, the distributed motor driving-braking torque distribution mode is selected; the second threshold value of the lateral instability level is determined by multiple simulations;

[0063] When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 1, and the lateral instability characterization factor S us When the lateral instability is greater than the second threshold value, a distributed motor drive-brake and hydraulic brake combined torque distribution mode is selected;

[0064] When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 0, and the lateral instability degree characterization factor S usWhen the lateral instability level is greater than the first threshold, a distributed motor drive and hydraulic brake combined torque distribution mode is selected;

[0065] When the driving motor state factor S m is equal to 0, and the power battery state factor S b When it is equal to 0, the hydraulic differential braking torque distribution mode is selected.

[0066] The beneficial effects of the present invention are:

[0067] The present invention uses the information collected by the driver-vehicle-road comprehensive information acquisition module to calculate the tire cornering stiffness mutation factor and selectively start the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm, and then accurately calculates the expected value of the control variable, and then uses the adaptive fuzzy integral sliding mode algorithm to calculate the target stability torque, and determines the torque distribution mode according to the numerical values ​​of the three factors and the torque distribution mode selection rules. The present invention can timely estimate the tire cornering stiffness according to the degree of mutation of the tire cornering stiffness, and ensure the lateral stability control effect of the electric vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0068] The present invention will be further described below in conjunction with the accompanying drawings:

[0069] Figure 1 This is a framework diagram of an intelligent vehicle stability control system based on tire cornering stiffness estimation proposed by the present invention. DETAILED DESCRIPTION

[0070] The present invention is further described in detail below based on the accompanying drawings and specific embodiments.

[0071] See also Figure 1 , the intelligent automobile stability control system based on tire cornering stiffness estimation described in the present invention comprises a driver-vehicle-road comprehensive information acquisition module, a control variable expected value calculation module, a target stability torque calculation module and a torque distribution mode selection module;

[0072] The driver-vehicle-road comprehensive information acquisition module is used to acquire the driver operation information, vehicle component use and driving status information and road condition information required to maintain the lateral stability of the vehicle; the driver operation information includes the steering wheel angle and the driver's grip on the steering wheel; the vehicle component use and driving status information includes tire aspect ratio, tire pressure, tire temperature, tire wear depth, tire lateral stiffness factory calibration value, battery charge state, motor power factor, longitudinal speed, vehicle mass, yaw angular velocity, center of mass sideslip angle and front wheel angle; the road condition information includes road adhesion coefficient, road curvature and road roughness;

[0073] The control variable expected value calculation module calculates the expected yaw angular velocity ω using the nonlinear two-degree-of-freedom vehicle model motion differential equation r_d and the desired sideslip angle β d The nonlinear two-degree-of-freedom vehicle model motion differential equation is based on the tire cornering stiffness mutation factor K tcs Establish: The tire cornering stiffness mutation factor K tcs , used to determine whether to start the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm; the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is designed with input state variables s and output action variables a as follows:

[0074]

[0075] In the formula, v x is the longitudinal speed; δ f is the front wheel turning angle; F z is the vertical load on the tire at the contact point; α is the tire slip angle; R tire is the coefficient reflecting the basic characteristics and usage status of the tire; μ is the road adhesion coefficient; C est is the estimated value of tire cornering stiffness;

[0076] The tire basic characteristics and usage status reflection coefficient R tire The calculation formula is:

[0077]

[0078] Wherein, P is the current tire pressure; P0 is the normal tire pressure, which is 2.2 bar; T is the current tire temperature; T0 is the normal tire temperature, which is 40°C; D is the current tire wear depth; D0 is the normal tire wear depth, which is 1.6 mm; τ is the characterization coefficient of the current tire structure type, if the current tire is a common bias tire, the value is 1, if the current tire is a radial tire, the value is 2; τ0 is the characterization coefficient of the common bias tire structure type, the value is 1; F is the current tire aspect ratio; F0 is the reference tire aspect ratio, which is 60%; κ1, κ2, κ3, κ4 and κ5 are weight coefficients, and the specific values ​​are determined by simulation and experiment;

[0079] The tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is designed with a reward function R as follows:

[0080]

[0081] Where C0 is the factory calibration value of the tire cornering stiffness; σ(s) is the standard deviation of the input state variable s; s i is the i-th component of the input state variable s; μi is the mean value of the i-th component of the input state variable s; σ i is the standard deviation of the i-th component of the input state variable s; p is the penalty term power, which is 2; λ1 and λ2 are weight coefficients, whose specific values ​​are determined through simulation and experiments;

[0082] The target stability torque calculation module is used to calculate the additional yaw torque required to maintain the lateral stability of the vehicle; the additional yaw torque is obtained based on an adaptive fuzzy integral sliding mode algorithm; the adaptive fuzzy integral sliding mode algorithm, wherein the input of the fuzzy algorithm is the vehicle operation feedback coefficient Q vof and the driver operation feedback coefficient Q dof , the output is the integral term coefficient c of the integral sliding surface;

[0083] The torque distribution mode selection module is based on the driving motor state factor S m , power battery state factor S b and the lateral instability characterization factor S us , and combined with the torque distribution method selection rules, determine the torque distribution mode; the torque distribution mode is specifically divided into five modes: distributed motor drive torque distribution mode, distributed motor drive-braking torque distribution mode, distributed motor drive-braking and hydraulic braking combined torque distribution mode, distributed motor drive and hydraulic braking combined torque distribution mode and hydraulic differential braking torque distribution mode.

[0084] The tire cornering stiffness mutation factor K tcs The branching factor F is used by the vehicle and components. vcu , vehicle driving state branch factor F vrs and vehicle driving condition branching factor F voc constitute;

[0085] The vehicle and components use branching factor F vcu The calculation formula is:

[0086]

[0087] In the formula, t v is the vehicle usage time; t t How long the tires last; w T and w P They are the vehicle usage time weight coefficient, tire usage time weight coefficient, current tire temperature weight coefficient and current tire pressure weight coefficient, and their specific values ​​are determined through simulation and experiments;

[0088] The vehicle driving state branch factor F vrs The calculation formula is:

[0089]

[0090] Where θ is the steering wheel angle; w v 、w θ and w α They are the vertical load weight coefficient at the tire contact point, the longitudinal vehicle speed weight coefficient, the steering wheel angle weight coefficient and the tire slip angle weight coefficient, and the specific values ​​are determined through simulation and experiments;

[0091] The vehicle driving condition branching factor F voc The calculation formula is:

[0092] F voc =w μ ·μ 2 +w ψ ·(+u 2 )+w u ·log(1+u)

[0093] Where μ is the road adhesion coefficient; ψ is the road curvature; u is the road roughness; w μ 、w ψ and w u They are the road adhesion coefficient weight coefficient, road curvature weight coefficient and road roughness weight coefficient, and their specific values ​​are determined through simulation and experiments;

[0094] The tire cornering stiffness mutation factor K tcs The calculation formula is:

[0095]

[0096] In the formula, w vcu 、w vrs and w voc They are the branch factor weight coefficients for vehicle and component use, vehicle driving state branch factor weight coefficients, and vehicle driving condition branch factor weight coefficients, and their specific values ​​are determined through simulation and testing;

[0097] The tire cornering stiffness mutation factor K tcs , used to determine whether to start the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm. The specific judgment rule is: when the tire cornering stiffness mutation factor K tcs When the tire cornering stiffness mutation factor K0 is greater than or equal to the tire cornering stiffness mutation threshold K0, the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is turned on to obtain the tire cornering stiffness that changes under the current working condition to improve the lateral stability control effect; when the tire cornering stiffness mutation factor K tcsWhen it is less than the tire cornering stiffness mutation threshold K0, the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is turned off, and the factory-calibrated tire cornering stiffness is used to reduce the workload of the vehicle central controller;

[0098] The specific expression formula of the nonlinear two-degree-of-freedom vehicle model motion differential equation is:

[0099]

[0100] In the formula, C f is the sum of the tire cornering stiffness of the two front wheels; C r is the sum of the tire cornering stiffness of the two rear wheels; ω r is the yaw rate; β is the sideslip angle of the center of mass; a and b are the distances from the center of mass to the front and rear axes respectively; I z is the moment of inertia of the vehicle around the z-axis; and are factory calibration values ​​of the tire cornering stiffness of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and The estimated values ​​of the tire cornering stiffness of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and The tire cornering stiffness mutation factors of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and are the tire cornering stiffness mutation thresholds of the left front wheel, right front wheel, left rear wheel and right rear wheel, all of which are 0.5;

[0101] The desired yaw rate ω r_d and the desired center of mass sideslip angle β d The calculation formula is:

[0102]

[0103] In the formula, is the vehicle stability factor.

[0104] The vehicle operation feedback coefficient Q vof The calculation formula is:

[0105]

[0106] Where D fl , D fr , D rl and D rrThe current tire wear depths of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; F x_fl 、F x_fr 、F x_rl and F x_rr are the longitudinal forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; F y_fl 、F y_fr 、F y_rl and F y_rr are the lateral forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; F z_fl 、F z_fr 、F z_rl 、F z_rr are the vertical forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; μ fl , μ fr , μ rl and μ rr are the road adhesion coefficients of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; k tire and k v is the weight coefficient, and its specific value is determined through simulation and experiments;

[0107] The driver operation feedback coefficient Q dof The calculation formula is:

[0108]

[0109] In the formula, F driver The driver's grip on the steering wheel; is the steering wheel angle change rate; w θ , and are the weight coefficients of steering wheel angle, steering wheel angle change rate and driver’s steering wheel grip force, and their specific values ​​are determined through simulation and experiment;

[0110] The specific design of the sliding surface of the adaptive fuzzy integral sliding mode algorithm is as follows:

[0111] s=ω r -ω r_d +(β-β d )+c∫[ω r -ω r_d +ρ(β-β d )]dt

[0112] Where c is the integral term coefficient of the integral sliding surface, and its specific value is determined based on the vehicle operation feedback coefficient Q vof and the driver operation feedback coefficient Q dof It is obtained through fuzzy reasoning; ρ is the weight coefficient of the control variable, and the specific value is determined through simulation and experiment;

[0113] Using the adaptive fuzzy integral sliding mode algorithm, the additional yaw moment is obtained as:

[0114]

[0115] Where ε and k are sliding reaching law coefficients, and their specific values ​​are obtained through multiple simulations; sat(s) is a saturation function used to weaken chattering.

[0116] The driving motor state factor S m The calculation formula is:

[0117]

[0118] Where PF is the motor power factor; PF max and PF min They are the maximum and minimum values ​​of the motor power factor when the motor is working normally, which are determined through motor testing; T m is the motor temperature; and They are the maximum and minimum values ​​of the motor temperature when the motor is working normally, which are determined through motor tests; H(x) is the Heaviside step function;

[0119] The driving motor state factor S b The calculation formula is:

[0120]

[0121] Where SOC is the battery state of charge; T b is the battery temperature; and They are the maximum and minimum values ​​of the battery temperature when the battery is working normally, which are determined through battery testing;

[0122] The lateral instability degree characterization factor S us The calculation formula is:

[0123]

[0124] In the formula, a y is the lateral acceleration; β max and They are the maximum values ​​of longitudinal speed, yaw angular velocity, sideslip angle at the center of mass and lateral acceleration when the vehicle is in laterally stable state, which are determined through multiple simulations; ε1, ε2, ε3 and ε4 are weight coefficients, and their specific values ​​are determined through simulation and experiments.

[0125] The torque distribution mode selection rule is specifically as follows:

[0126] When the driving motor state factor S m is equal to 1, and the lateral instability degree characterization factor S us When the lateral instability level is less than a first threshold, the distributed motor drive torque distribution mode is selected; the first threshold of the lateral instability level is determined by multiple simulations;

[0127] When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 1, and the lateral instability characterization factor S us When the lateral instability level is within the range of the first threshold value and the second threshold value, the distributed motor driving-braking torque distribution mode is selected; the second threshold value of the lateral instability level is determined by multiple simulations;

[0128] When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 1, and the lateral instability characterization factor S us When the lateral instability is greater than the second threshold value, a distributed motor drive-brake and hydraulic brake combined torque distribution mode is selected;

[0129] When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 0, and the lateral instability degree characterization factor S us When the lateral instability level is greater than the first threshold, a distributed motor drive and hydraulic brake combined torque distribution mode is selected;

[0130] When the driving motor state factor S m is equal to 0, and the power battery state factor S b When it is equal to 0, the hydraulic differential braking torque distribution mode is selected.

Claims

1. An intelligent vehicle stability control system based on tire cornering stiffness estimation, characterized in that: The system is designed for electric vehicles driven by four-wheel hub motors and includes a driver-vehicle-road comprehensive information acquisition module, a control variable expected value calculation module, a target stability torque calculation module, and a torque distribution mode selection module; The driver-vehicle-road comprehensive information acquisition module is used to acquire the driver operation information, vehicle component use and driving status information and road condition information required to maintain the lateral stability of the vehicle; the driver operation information includes the steering wheel angle and the driver's grip on the steering wheel; the vehicle component use and driving status information includes tire aspect ratio, tire pressure, tire temperature, tire wear depth, tire lateral stiffness factory calibration value, battery charge state, motor power factor, longitudinal speed, vehicle mass, yaw angular velocity, center of mass sideslip angle and front wheel angle; the road condition information includes road adhesion coefficient, road curvature and road roughness; The control variable expected value calculation module calculates the expected yaw angular velocity ω using the nonlinear two-degree-of-freedom vehicle model motion differential equation r_d and the desired sideslip angle β d The nonlinear two-degree-of-freedom vehicle model motion differential equation is based on the tire cornering stiffness mutation factor K tcs Establish: The tire cornering stiffness mutation factor K tcs , used to determine whether to start the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm; the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is designed with input state variables s and output action variables a as follows: In the formula, v x is the longitudinal speed; δ f is the front wheel turning angle; F z is the vertical load on the tire at the contact point; α is the tire slip angle; R tire is the coefficient reflecting the basic characteristics and usage status of the tire; μ is the road adhesion coefficient; C est is the estimated value of tire cornering stiffness; The tire basic characteristics and usage status reflection coefficient R tire The calculation formula is: Wherein, P is the current tire pressure; P0 is the normal tire pressure, which is 2.2 bar; T is the current tire temperature; T0 is the normal tire temperature, which is 40°C; D is the current tire wear depth; D0 is the normal tire wear depth, which is 1.6 mm; τ is the characterization coefficient of the current tire structure type, if the current tire is a common bias tire, the value is 1, if the current tire is a radial tire, the value is 2; τ0 is the characterization coefficient of the common bias tire structure type, the value is 1; F is the current tire aspect ratio; F0 is the reference tire aspect ratio, which is 60%; κ1, κ2, κ3, κ4 and κ5 are weight coefficients, and the specific values ​​are determined by simulation and experiment; The tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is designed with a reward function R as follows: Where C0 is the factory calibration value of the tire cornering stiffness; σ(s) is the standard deviation of the input state variable s; s i is the i-th component of the input state variable s; μ i is the mean value of the i-th component of the input state variable s; σ i is the standard deviation of the i-th component of the input state variable s; p is the penalty term power, which is 2; λ1 and λ2 are weight coefficients, whose specific values ​​are determined through simulation and experiments; The target stability torque calculation module is used to calculate the additional yaw torque required to maintain the lateral stability of the vehicle; the additional yaw torque is obtained based on an adaptive fuzzy integral sliding mode algorithm; the adaptive fuzzy integral sliding mode algorithm, wherein the input of the fuzzy algorithm is the vehicle operation feedback coefficient Q vof and the driver operation feedback coefficient Q dof , the output is the integral term coefficient c of the integral sliding surface; The torque distribution mode selection module is based on the driving motor state factor S m , power battery state factor S b and the lateral instability characterization factor S us , and combined with the torque distribution method selection rules, determine the torque distribution mode; the torque distribution mode is specifically divided into five modes: distributed motor drive torque distribution mode, distributed motor drive-braking torque distribution mode, distributed motor drive-braking and hydraulic braking combined torque distribution mode, distributed motor drive and hydraulic braking combined torque distribution mode and hydraulic differential braking torque distribution mode.

2. The intelligent vehicle stability control system based on tire cornering stiffness estimation according to claim 1, characterized in that: The tire cornering stiffness mutation factor K tcs The branching factor F is used by the vehicle and components. vcu , vehicle driving state branch factor F vrs and vehicle driving condition branching factor F voc constitute; The vehicle and components use branching factor F vcu The calculation formula is: Where, t v is the vehicle usage time; t t How long the tires last; w T and w P They are the vehicle usage time weight coefficient, tire usage time weight coefficient, current tire temperature weight coefficient and current tire pressure weight coefficient, and their specific values ​​are determined through simulation and experiments; The vehicle driving state branch factor F vrs The calculation formula is: Where θ is the steering wheel angle; w v 、w θ and w α They are the vertical load weight coefficient at the tire contact point, the longitudinal vehicle speed weight coefficient, the steering wheel angle weight coefficient and the tire slip angle weight coefficient, and the specific values ​​are determined through simulation and experiments; The vehicle driving condition branching factor F voc The calculation formula is: F voc =w μ ·μ 2 +w ψ ·(ψ+u 2 )+w u ·log(1+u) Where μ is the road adhesion coefficient; ψ is the road curvature; u is the road roughness; w μ 、w ψ and w u They are the road adhesion coefficient weight coefficient, road curvature weight coefficient and road roughness weight coefficient, and their specific values ​​are determined through simulation and experiments; The tire cornering stiffness mutation factor K tcs The calculation formula is: In the formula, w vcu 、w vrs and w voc They are the branch factor weight coefficients for vehicle and component use, vehicle driving state branch factor weight coefficients, and vehicle driving condition branch factor weight coefficients, and their specific values ​​are determined through simulation and testing; The tire cornering stiffness mutation factor K tcs , used to determine whether to start the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm. The specific judgment rule is: when the tire cornering stiffness mutation factor K tcs When the tire cornering stiffness mutation factor K0 is greater than or equal to the tire cornering stiffness mutation threshold K0, the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is turned on to obtain the tire cornering stiffness that changes under the current working condition to improve the lateral stability control effect; when the tire cornering stiffness mutation factor K tcs When it is less than the tire cornering stiffness mutation threshold K0, the tire cornering stiffness dynamic estimator based on the deep deterministic policy gradient algorithm is turned off, and the factory-calibrated tire cornering stiffness is used to reduce the workload of the vehicle central controller; The specific expression formula of the nonlinear two-degree-of-freedom vehicle model motion differential equation is: In the formula, C f is the sum of the tire cornering stiffness of the two front wheels; C r is the sum of the tire cornering stiffness of the two rear wheels; ω r is the yaw angular velocity; β is the sideslip angle of the center of mass; a and b are the distances from the center of mass to the front and rear axes respectively; I z is the moment of inertia of the vehicle around the z-axis; and are factory calibration values ​​of the tire cornering stiffness of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and The estimated values ​​of the tire cornering stiffness of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and The tire cornering stiffness mutation factors of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; and are the tire cornering stiffness mutation thresholds of the left front wheel, right front wheel, left rear wheel and right rear wheel, all of which are 0.5; The desired yaw rate ω r_d and the desired center of mass sideslip angle β d The calculation formula is: In the formula, is the vehicle stability factor.

3. The intelligent vehicle stability control system based on tire cornering stiffness estimation according to claim 1, characterized in that: The vehicle operation feedback coefficient Q vof The calculation formula is: Where D f1 , D fr , D rl and D rr The current tire wear depths of the left front wheel, the right front wheel, the left rear wheel and the right rear wheel respectively; F x_f1 、F x_fr 、F x_rl and F x_rr are the longitudinal forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; F y_f1 、F y_fr 、F y_rl and F y_rr are the lateral forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; F z_fl 、F z_fr 、F z_r1 、F z_rr are the vertical forces of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; μ fl , μ fr , μ r1 and μ rr are the road adhesion coefficients of the left front wheel, right front wheel, left rear wheel and right rear wheel respectively; k tire and k v is the weight coefficient, and its specific value is determined through simulation and experiment; The driver operation feedback coefficient Q dof The calculation formula is: In the formula, F driver The driver's grip on the steering wheel; is the steering wheel angle change rate; w θ , and are the weight coefficients of steering wheel angle, steering wheel angle change rate and driver’s steering wheel grip force, and their specific values ​​are determined through simulation and experiment; The specific design of the sliding surface of the adaptive fuzzy integral sliding mode algorithm is as follows: s = ω r -oh r_d +(β-β d )+c∫[ω r -oh r_d +ρ(β-β d )]dt Where c is the integral term coefficient of the integral sliding surface, and its specific value is determined based on the vehicle operation feedback coefficient Q vof and the driver operation feedback coefficient Q dof Obtained through fuzzy reasoning; ρ is the weight coefficient of the control variable, and its specific value is determined through simulation and experiment; Using the adaptive fuzzy integral sliding mode algorithm, the additional yaw moment is obtained as: Where ε and k are sliding reaching law coefficients, and their specific values ​​are obtained through multiple simulations; sat(s) is a saturation function used to weaken chattering.

4. The intelligent vehicle stability control system based on tire cornering stiffness estimation according to claim 1, characterized in that: The driving motor state factor S m The calculation formula is: Where PF is the motor power factor; PF max and PF min They are the maximum and minimum values ​​of the motor power factor when the motor is working normally, which are determined through motor testing; T m is the motor temperature; and They are the maximum and minimum values ​​of the motor temperature when the motor is working normally, which are determined through motor tests; H(x) is the Heaviside step function; The driving motor state factor S b The calculation formula is: Where SOC is the battery state of charge; T b is the battery temperature; and They are the maximum and minimum values ​​of the battery temperature when the battery is working normally, which are determined through battery testing; The lateral instability degree characterization factor S us The calculation formula is: In the formula, a y is the lateral acceleration; β max and They are the maximum values ​​of longitudinal speed, yaw angular velocity, sideslip angle at the center of mass and lateral acceleration when the vehicle is in laterally stable state, which are determined through multiple simulations; ε1, ε2, ε3 and ε4 are weight coefficients, and their specific values ​​are determined through simulation and experiments.

5. The intelligent vehicle stability control system based on tire cornering stiffness estimation according to claim 1, characterized in that: The torque distribution mode selection rule is specifically as follows: When the driving motor state factor S m is equal to 1, and the lateral instability degree characterization factor S us When the lateral instability level is less than a first threshold, the distributed motor drive torque distribution mode is selected; the first threshold of the lateral instability level is determined by multiple simulations; When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 1, and the lateral instability characterization factor S us When the lateral instability level is within the range of the first threshold value and the second threshold value, the distributed motor driving-braking torque distribution mode is selected; the second threshold value of the lateral instability level is determined by multiple simulations; When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 1, and the lateral instability characterization factor S us When the lateral instability is greater than the second threshold value, a distributed motor drive-brake and hydraulic brake combined torque distribution mode is selected; When the driving motor state factor S m is equal to 1, and the power battery state factor S b is equal to 0, and the lateral instability degree characterization factor S us When the lateral instability level is greater than the first threshold, a distributed motor drive and hydraulic brake combined torque distribution mode is selected; When the driving motor state factor S m is equal to 0, and the power battery state factor S b When it is equal to 0, the hydraulic differential braking torque distribution mode is selected.

Images