Method and system for estimating road adhesion coefficient of electric vehicle

Abstract

The invention discloses a road adhesion coefficient estimation method and system for an electric vehicle, and relates to the technical field of electric vehicles, and the method comprises the steps: constructing a seven-degree-of-freedom model of a distributed driving electric vehicle; based on a Dugoff tire model, determining a nonlinear relationship among a vehicle motion state, a road adhesion coefficient and tire force, and fitting to obtain normalized tire longitudinal and lateral forces; judging a current working condition according to a vehicle driving state, constructing a longitudinal working condition model and a transverse and longitudinal coupling working condition model, calculating model matching probabilities of different working conditions based on a Bayesian method, and estimating a road adhesion coefficient; introducing an adaptive square root cubature Kalman filtering algorithm, and designing an adaptive strategy based on observed quantity residual sum of squares and a sliding window method to estimate a road adhesion coefficient; and model estimation results are fused through weighted average to obtain a road adhesion coefficient estimation value, so that the road adhesion coefficient can be accurately estimated according to the dynamic information of the vehicle under different driving conditions, and the stability and safety of the vehicle are improved.

Description

Technical Field

[0001] The invention relates to the technical field of electric vehicles, and in particular to a method and a system for estimating a road adhesion coefficient of an electric vehicle. Background Art

[0002] With the continuous advancement of electric vehicle technology, especially the rapid development of distributed drive electric vehicles, the power distribution and control capabilities of vehicles have been significantly improved. This type of electric vehicle uses multiple motors to drive each wheel separately, which has better power distribution and vehicle control capabilities compared to the traditional central drive method. The active safety control strategy of the vehicle needs to adjust the controller parameters in real time according to different road surfaces to improve the stability of the vehicle. The road adhesion coefficient determines the size of the tire force, which directly affects the driving performance and stability of the vehicle. Therefore, the accurate identification and estimation of the road adhesion coefficient has gradually attracted the attention of the electric vehicle industry.

[0003] In the prior art, the cause-based method directly collects data on factors that affect the road adhesion coefficient, such as road gloss, roughness, etc., through sensors, and then infers the result. However, for the same type of road surface, the road adhesion coefficient is not constant, and the sensor is expensive, which increases the manufacturing cost of the whole vehicle. In addition, this method is easily affected by the quality of training data, which makes it limited in accurately identifying the road adhesion coefficient. The effect-based method estimates the road adhesion coefficient based on the vehicle's response to road excitation, and requires the establishment of an accurate vehicle dynamics model and tire model. Although low-cost on-board sensors are used to collect vehicle state parameters, the filter estimation algorithm still has great limitations in dealing with nonlinearity, dealing with noise, and dealing with uncertainty. For example, although the unscented Kalman filter and the volumetric Kalman filter have advantages, in practical applications, they still need to be adaptively processed for noise changes to improve estimation accuracy.

[0004] Therefore, how to accurately estimate the road adhesion coefficient based on the vehicle dynamic information under different driving conditions and improve the stability and safety of the vehicle has become a problem that needs to be solved urgently. Summary of the invention

[0005] In order to solve the above problems, the present invention proposes a method and system for estimating the road adhesion coefficient of an electric vehicle, which solves the problems of low accuracy and poor adaptability of existing road adhesion coefficient estimation methods. The method can accurately estimate the road adhesion coefficient based on the vehicle dynamic information under different driving conditions, thereby improving the stability and safety of the vehicle.

[0006] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0007] A method for estimating a road adhesion coefficient of an electric vehicle according to the present invention comprises:

[0008] Based on the vehicle kinematic equations and wheel rotation equations, a 7-DOF model of distributed drive electric vehicles is constructed, including longitudinal, lateral, yaw and rotational motion of the four wheels;

[0009] Based on the Dugoff tire model, the nonlinear relationship between the vehicle motion state, road adhesion coefficient and tire force is determined, and the normalized tire longitudinal and lateral forces are obtained by fitting the vehicle motion state.

[0010] The current working condition is determined according to the driving state of the vehicle, and a longitudinal working condition model and a transverse and longitudinal coupling working condition model are constructed for the straight working condition and the turning working condition respectively, and the corresponding state equation and observation equation are established. The model matching probability under different working conditions is calculated based on the Bayesian method, and the corresponding model is dynamically selected according to the model matching probability to estimate the road adhesion coefficient;

[0011] Introduce an adaptive square root cubature Kalman filter algorithm to estimate the road adhesion coefficient, construct a square root cubature Kalman filter algorithm, including state prediction and measurement update, design an adaptive strategy based on the sum of squares of observation residuals and the sliding window method, calculate the residuals and update the noise covariance matrix, update the measurement noise covariance matrix in real time, and keep the process noise covariance matrix unchanged;

[0012] The estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model are fused through the weighted average method to obtain the fused estimated value of the road adhesion coefficient.

[0013] A further improvement of the present invention is that: based on the vehicle kinematic equation and the wheel rotation equation, a 7-degree-of-freedom model of a distributed drive electric vehicle is constructed, including longitudinal, lateral, yaw and rotational motion of four wheels, specifically including:

[0014] The 7-DOF model of the distributed drive electric vehicle includes longitudinal, lateral, yaw and rotational motion of the four wheels;

[0015] The vertical and lateral motions of the vehicle are ignored, and the vehicle is traveling on a horizontal road; the influence of the suspension on the vehicle is ignored, and the entire vehicle body is regarded as a rigid body; the influence of air resistance is ignored; the influence of the steering system is ignored, and the steering wheel angle is directly converted into the front wheel angle input. At this time, the vehicle kinematic equation is:

[0016]

[0017] For the rotational motion of a single wheel, according to the force on the wheel, the wheel rotation equation is:

[0018] I ω ω i =T di -Fxi R e (4)

[0019] In the formula, v x is the longitudinal speed of the vehicle; v y is the lateral speed of the vehicle; a x is the longitudinal acceleration of the vehicle; a y is the lateral acceleration of the vehicle; γ is the yaw rate; m is the vehicle mass; F xi is the tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi is the tire lateral force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; a and b are the distances from the center of mass to the front and rear axles respectively; ω i is the wheel speed, i.e. the angular acceleration of the wheel; δ is the front wheel turning angle; d f ,d r I is the front and rear wheelbase; z is the moment of inertia of the vehicle around the z axis; I ω is the moment of inertia of the wheel around the axle; T di is the driving torque; R e is the wheel radius; sin is the sine function; cos is the cosine function.

[0020] A further improvement of the present invention is that: the nonlinear relationship between the vehicle motion state, the road adhesion coefficient and the tire force is determined based on the Dugoff tire model, and the normalized tire longitudinal and lateral forces are obtained by fitting the vehicle motion state, specifically including:

[0021] Based on the Dugoff tire model, the nonlinear relationship between the vehicle motion state, road adhesion coefficient and tire force is determined, and the road adhesion coefficient variable is separated. The normalized tire longitudinal and lateral forces are obtained by fitting the vehicle motion state, that is:

[0022]

[0023] In the formula, F x 、F y F is the longitudinal force and lateral force of the tire respectively; x 0 、F y 0 are the normalized tire longitudinal force and lateral force respectively; C x , C y are the longitudinal stiffness and lateral stiffness of the tire respectively; P(L) is the function of tire force used to correct the nonlinear interval; L is the tire working state parameter, L<1 means the tire is in the nonlinear region, otherwise it is in the linear region; κ and α are the slip rate and sideslip angle respectively; ε is the speed influence factor; μ is the road adhesion coefficient; Fz is the vertical load of the tire; v x is the longitudinal speed of the vehicle; tan is the tangent function.

[0024] A further improvement of the present invention is that: the current working condition is judged according to the driving state of the vehicle, a longitudinal working condition model and a transverse and longitudinal coupling working condition model are respectively constructed for the straight working condition and the turning working condition, the corresponding state equation and observation equation are established, and the model matching probability under different working conditions is calculated based on the Bayesian method, and the corresponding model is dynamically selected according to the model matching probability to estimate the road adhesion coefficient, which specifically includes:

[0025] For the nonlinear system state estimation model composed of state variables x(t) and observation variables z(t) with Gaussian white noise w(t) and v(t), the road adhesion coefficient state equation and measurement equation are established as follows:

[0026]

[0027] Where f() and h() are the system state function and measurement function respectively; w(t) is the process noise at time t; v(t) is the measurement noise at time t; x(t) is the state variable at time t; z(t) is the observed variable at time t;

[0028] The road adhesion coefficient estimation algorithm based on Kalman filter takes the current road adhesion coefficient μ of each wheel as i As state variable x = (μ1, μ2, μ3, μ4) T ; For different driving conditions, different sensor signals are selected as observed variables. For straight-line driving conditions, the longitudinal acceleration a of the vehicle is selected. x and the angular acceleration of the wheel ω i As the observed variable z = (a x ,ω1,ω2,ω3,ω4) T ; For cornering conditions, select the vehicle's longitudinal acceleration a x , lateral acceleration a y and yaw rate γ as observed variables z = (a x ,a y ,γ) T , the sampling time of the Kalman filter is set to Δt, and the state space equation established after the system is discretized is as follows:

[0029]

[0030] In the formula, x(k+1) represents the state variable at time k+1; x(k) represents the state variable at time k; w(k) is the process noise at time k;

[0031] For the estimation of the road adhesion coefficient under different driving conditions, the conditions are divided into longitudinal conditions and transverse and longitudinal coupled conditions according to the driving state of the vehicle. The Bayesian method is used for model matching, the model matching probability under different conditions is calculated, and the corresponding model is dynamically selected for road adhesion coefficient estimation based on the model matching probability.

[0032] A further improvement of the present invention is that when the Bayesian method is used to calculate the model matching probability under different working conditions, the lateral acceleration and yaw angular velocity are first used as input features, the prior probability is set according to the driving condition of the vehicle, and the lateral acceleration and yaw angular velocity are set to obey the Gaussian distribution in each state, and a likelihood function is established. Then, the posterior probability in each state is calculated based on the Bayesian theorem, and the corresponding model is dynamically selected to estimate the road adhesion coefficient based on the maximum a posteriori probability criterion.

[0033] A further improvement of the present invention is that the longitudinal working condition model is used for estimating the longitudinal working condition of the vehicle, ignoring the lateral force in the observation equation, and introducing the angular acceleration ω of the four wheels i , set the observed variable of the longitudinal operating model to z x =(a x ,ω1,ω2,ω3,ω4) T , the observation equation of the longitudinal working condition model is:

[0034]

[0035] In the formula, z x (k) is the observed variable of the longitudinal operating model at time k; F xi 0 is the normalized tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi 0 is the normalized tire lateral force, where i = 1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; δ is the front wheel turning angle; m is the vehicle mass; R e is the wheel radius; I ω is the moment of inertia of the wheel around the axle; x(k) is the state variable at time k; T di is the driving torque; I ω is the moment of inertia of the wheel around the axle; v(k) is the measurement noise at time k; ψ is the longitudinal working condition model parameter, and ψ=(0,1,1,1,1) T ; cos is the cosine function.

[0036] A further improvement of the present invention is that: the lateral and longitudinal coupling working condition model is used to estimate the longitudinal working condition of the vehicle, and the derivatives of the longitudinal acceleration, lateral acceleration and yaw rate of the vehicle are selected as observation variables in the observation equation to obtain zxy =(a x ,a y ,γ) T ;

[0037] The observation equation of the transverse and longitudinal coupling model is:

[0038]

[0039] In the above formula, the parameters of the observation matrix are:

[0040]

[0041] Where zxy(k) is the observed variable of the lateral and longitudinal coupling model at time k; F xi 0 is the normalized tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi 0 is the normalized tire lateral force, where i = 1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; δ is the front wheel turning angle; m is the vehicle mass; d f ,d r I is the front and rear wheelbase; z is the moment of inertia of the vehicle around the z-axis; a and b are the distances from the center of mass to the front and rear axes, respectively; H(3,1), H(3,2), H(3,3), H(3,4) are the observation matrix parameters; sin is the sine function; cos is the cosine function; d f ,d r Front and rear wheelbase

[0042] The further improvement of the present invention is that: the adaptive square root volumetric Kalman filter algorithm is introduced to estimate the road adhesion coefficient, a square root volumetric Kalman filter algorithm is constructed, including state prediction and measurement update, an adaptive strategy based on the observed residual square sum and sliding window method is designed, residuals are calculated and the noise covariance matrix is ​​updated, and the measurement noise covariance matrix is ​​updated in real time while keeping the process noise covariance matrix unchanged, specifically including:

[0043] According to the optimal estimate of the filter k-1 at the previous moment And the square root of the corresponding error covariance matrix S(k-1|k-1), the state estimate at time k is solved through time update and measurement update and the square root of the error covariance matrix S(k-1|k-1);

[0044] Combined with the sliding window method, the objective function based on the observation error is obtained as the iterative judgment condition. The corresponding calculation formula is as follows:

[0045]

[0046] Where J is the objective function based on observation error; n is the total number of sliding windows; W is the size of the sliding window; e(k) is the observation error at time k; e T (k) is the matrix transpose of the observation error at time k; η is the output value error weight matrix;

[0047] The calculation formula of the observation error e(k) at time k is as follows:

[0048]

[0049] Where, e(k) is the observation error at time k; z(k) is the observation variable at time k; is the optimal estimate of the observed variables from the previous time k-1 to time k;

[0050] Combined with the order of magnitude differences of different output values, the output value error weight matrix η is introduced:

[0051] In cornering conditions,

[0052]

[0053] Where η1, η2, η3 are the error weights of the output values;

[0054] In straight driving condition,

[0055]

[0056] Where η1 and η2 are the error weights of the output values;

[0057] By updating the measurement noise covariance matrix in real time and keeping the process noise covariance matrix unchanged, the measurement noise is updated as follows by combining the filtering process of the adaptive square root cubature Kalman filter algorithm and the adaptive strategy based on the sum of squares of the observed residuals and the sliding window method:

[0058]

[0059] Where H(k) is the residual of the observation at time k; e(k) is the observation error at time k; diag(e(k)) is the diagonal matrix of the observation error at time k; W is the size of the sliding window; is the adaptive smoothing factor; R(k) is the measurement noise covariance matrix at time k; R(k-1) is the measurement noise covariance matrix at the previous time k-1.

[0060] A road adhesion coefficient estimation system for an electric vehicle of the present invention comprises:

[0061] A construction module is used to construct a 7-DOF model of a distributed drive electric vehicle based on the vehicle kinematic equations and wheel rotation equations, including longitudinal, lateral, yaw and rotational motion of the four wheels;

[0062] A fitting module is used to determine the nonlinear relationship between the vehicle motion state, the road adhesion coefficient and the tire force based on the Dugoff tire model, and obtain the normalized tire longitudinal and lateral forces by fitting the vehicle motion state;

[0063] An estimation module is used to determine the current working condition according to the driving state of the vehicle, construct a longitudinal working condition model and a transverse and longitudinal coupling working condition model for the straight working condition and the turning working condition respectively, establish the corresponding state equation and observation equation, and calculate the model matching probability under different working conditions based on the Bayesian method, and dynamically select the corresponding model according to the model matching probability to estimate the road adhesion coefficient;

[0064] The update module is used to introduce the adaptive square root volumetric Kalman filter algorithm to estimate the road adhesion coefficient, construct the square root volumetric Kalman filter algorithm, including state prediction and measurement update, design an adaptive strategy based on the sum of squares of observed residuals and the sliding window method, calculate the residuals and update the noise covariance matrix, update the measurement noise covariance matrix in real time, and keep the process noise covariance matrix unchanged;

[0065] The fusion module is used to fuse the estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model through a weighted average method to obtain a fused estimated value of the road adhesion coefficient.

[0066] The beneficial effects of the present invention are as follows: the present invention proposes a method and system for estimating the road adhesion coefficient of an electric vehicle, and constructs a 7-degree-of-freedom model of a distributed drive electric vehicle based on the vehicle kinematic equation and the wheel rotation equation, including longitudinal, lateral, yaw and rotational motions of four wheels; determines the nonlinear relationship between the vehicle motion state, the road adhesion coefficient and the tire force based on the Dugoff tire model, and obtains the normalized tire longitudinal and lateral forces through vehicle motion state fitting; judges the current working condition according to the vehicle driving state, constructs a longitudinal working condition model and a transverse and longitudinal coupling working condition model for straight-line working conditions and turning working conditions respectively, establishes corresponding state equations and observation equations, calculates the model matching probability under different working conditions based on the Bayesian method, and dynamically selects the corresponding model to estimate the road adhesion coefficient according to the model matching probability; introduces an adaptive square root volumetric Kalman filter algorithm to estimate the road adhesion coefficient The invention discloses a method for estimating a road surface adhesion coefficient by using a weighted average method, and constructing a square root volumetric Kalman filter algorithm, including state prediction and measurement update, and designing an adaptive strategy based on the residual square sum of observed quantities and a sliding window method, calculating the residual and updating the noise covariance matrix, and updating the measurement noise covariance matrix in real time, while keeping the process noise covariance matrix unchanged; the estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model are fused by a weighted average method to obtain the fused road adhesion coefficient estimation value, so that dynamic switching can be performed according to different driving conditions, thereby improving the adaptability and accuracy of the estimation; and by introducing an adaptive strategy and a Bayesian classification method, the stability and accuracy of the estimation are further improved, thereby providing support for the active safety control of the vehicle; at the same time, the method of the invention has broad application prospects, and can be applied to the fields of safety control systems, intelligent driving assistance systems, etc. of electric vehicles, so as to provide guarantee for the safe driving of electric vehicles. BRIEF DESCRIPTION OF THE DRAWINGS

[0067] Figure 1 A schematic flow chart of a method for estimating a road adhesion coefficient of an electric vehicle provided by an embodiment of the present invention;

[0068] Figure 2 A schematic diagram of the structure of a road adhesion coefficient estimation system for an electric vehicle provided by an embodiment of the present invention;

[0069] Figure 3 A schematic diagram of the hardware structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0071] See also Figure 1 , is a flow chart of a method for estimating a road adhesion coefficient of an electric vehicle provided by an embodiment of the present invention, Figure 1 The execution subject of the method shown may be a software and / or hardware device. The execution subject of the present application may include but is not limited to at least one of the following: user equipment, network equipment, etc. Among them, user equipment may include but is not limited to computers, smart phones, personal digital assistants (Personal Digital Assistant, referred to as: PDA) and the electronic devices mentioned above. Network equipment may include but is not limited to a single network server, a server group consisting of multiple network servers, or a cloud based on cloud computing consisting of a large number of computers or network servers, wherein cloud computing is a type of distributed computing, a super virtual computer composed of a group of loosely coupled computers. This embodiment does not limit this. It includes steps S1 to S5, as follows:

[0072] S1, based on the vehicle kinematic equations and wheel rotation equations, a 7-DOF model of the distributed drive electric vehicle is constructed, including longitudinal, lateral, yaw and rotational motion of the four wheels;

[0073] S2, based on the Dugoff tire model, determines the nonlinear relationship between the vehicle motion state, road adhesion coefficient and tire force, and obtains the normalized tire longitudinal and lateral forces by fitting the vehicle motion state;

[0074] S3, judging the current working condition according to the driving state of the vehicle, constructing a longitudinal working condition model and a transverse and longitudinal coupling working condition model for the straight working condition and the turning working condition respectively, establishing the corresponding state equation and observation equation, and calculating the model matching probability under different working conditions based on the Bayesian method, and dynamically selecting the corresponding model according to the model matching probability to estimate the road adhesion coefficient;

[0075] S4, introduces the adaptive square root volumetric Kalman filter algorithm to estimate the road adhesion coefficient, constructs the square root volumetric Kalman filter algorithm, including state prediction and measurement update, designs an adaptive strategy based on the sum of squares of observation residuals and the sliding window method, calculates the residuals and updates the noise covariance matrix, updates the measurement noise covariance matrix in real time, and keeps the process noise covariance matrix unchanged;

[0076] S5, the estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model are fused by a weighted average method to obtain a fused estimated value of the road adhesion coefficient.

[0077] It can be understood that, firstly, a 7-DOF model can be constructed based on the vehicle kinematic equations and wheel rotation equations. The model includes the vehicle's longitudinal displacement, lateral displacement, yaw angle motion, and rotational motion of four independent wheels, thereby simulating the vehicle's motion state in multi-dimensional space, which is convenient for subsequent vehicle dynamics analysis and control strategy design.

[0078] In order to deeply analyze the interaction between the vehicle and the road, the Dugoff tire model can be used to accurately reflect the nonlinear relationship between the vehicle's motion state, the road adhesion coefficient, and the tire force. By real-time monitoring and fitting the vehicle's motion state, the normalized tire longitudinal force and lateral force can be calculated, and then the vehicle's stability and handling performance can be evaluated.

[0079] According to the current driving state of the vehicle, a working condition identification strategy can be formulated to divide the driving scene into straight-line driving conditions and turning conditions. For these two conditions, a longitudinal working condition model and a transverse and longitudinal coupling working condition model are constructed respectively, and the corresponding state equation and observation equation are established. The Bayesian method can be used to calculate the probability of model matching under different working conditions, and dynamically select the most appropriate model to estimate the road adhesion coefficient, thereby effectively improving the accuracy and adaptability of the estimation.

[0080] In order to further improve the accuracy and robustness of the estimation of the road adhesion coefficient, an adaptive square root cubature Kalman filter algorithm can be introduced. This algorithm includes two core steps: state prediction and measurement update, and can efficiently handle the state estimation problem in nonlinear dynamic systems. In addition, an adaptive strategy based on the sum of squares of the observed residuals and the sliding window method can be designed to calculate the residuals in real time and update the noise covariance matrix, thereby accurately capturing and dynamically adjusting the measurement noise. In this process, the process noise covariance matrix remains unchanged, ensuring the stability and reliability of the algorithm.

[0081] In order to obtain an accurate estimate of the road adhesion coefficient, a weighted average method can be used to fuse the estimation results based on the longitudinal operating condition model and the transverse and longitudinal coupling operating condition model, so as to fully utilize the advantages of different models under different operating conditions. Through reasonable weight distribution, the estimation results can be optimized and complemented, and finally the fused estimate of the road adhesion coefficient can be obtained to provide support for safe driving and intelligent control of the vehicle.

[0082] Based on the vehicle kinematic equation and wheel rotation equation, a 7-DOF model of a distributed drive electric vehicle is constructed, including longitudinal, lateral, yaw and rotational motion of four wheels, specifically including:

[0083] The 7-DOF model of the distributed drive electric vehicle includes longitudinal, lateral, yaw and rotational motion of the four wheels;

[0084] The vertical and lateral motions of the vehicle are ignored, and the vehicle is traveling on a horizontal road; the influence of the suspension on the vehicle is ignored, and the entire vehicle body is regarded as a rigid body; the influence of air resistance is ignored; the influence of the steering system is ignored, and the steering wheel angle is directly converted into the front wheel angle input. At this time, the vehicle kinematic equation is:

[0085]

[0086]

[0087] For the rotational motion of a single wheel, according to the force on the wheel, the wheel rotation equation is:

[0088] I ω ω i =T di -F xi R e (4)

[0089] In the formula, v x is the longitudinal speed of the vehicle; v y is the lateral speed of the vehicle; a x is the longitudinal acceleration of the vehicle; a y is the lateral acceleration of the vehicle; γ is the yaw rate; m is the vehicle mass; F xi is the tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi is the tire lateral force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; a and b are the distances from the center of mass to the front and rear axles respectively; ω i is the wheel speed, i.e. the angular acceleration of the wheel; δ is the front wheel turning angle; d f ,d r I is the front and rear wheelbase; z is the moment of inertia of the vehicle around the z axis; I ω is the moment of inertia of the wheel around the axle; T di is the driving torque; R e is the wheel radius; sin is the sine function; cos is the cosine function.

[0090] The method of determining the nonlinear relationship between the vehicle motion state, the road adhesion coefficient and the tire force based on the Dugoff tire model, and obtaining the normalized tire longitudinal and lateral forces by fitting the vehicle motion state, specifically includes:

[0091] Based on the Dugoff tire model, the nonlinear relationship between the vehicle motion state, road adhesion coefficient and tire force is determined, and the road adhesion coefficient variable is separated. The normalized tire longitudinal and lateral forces are obtained by fitting the vehicle motion state, that is:

[0092]

[0093] In the formula, F x 、F y F is the longitudinal force and lateral force of the tire respectively; x 0 、F y 0 are the normalized tire longitudinal force and lateral force respectively; C x , C y are the longitudinal stiffness and lateral stiffness of the tire respectively; P(L) is the function of tire force used to correct the nonlinear interval; L is the tire working state parameter, L<1 means the tire is in the nonlinear region, otherwise it is in the linear region; κ and α are the slip rate and sideslip angle respectively; ε is the speed influence factor; μ is the road adhesion coefficient; F z is the vertical load of the tire; v x is the longitudinal speed of the vehicle; tan is the tangent function.

[0094] The current working condition is judged according to the driving state of the vehicle, a longitudinal working condition model and a transverse and longitudinal coupling working condition model are constructed for the straight working condition and the turning working condition respectively, the corresponding state equation and observation equation are established, and the model matching probability under different working conditions is calculated based on the Bayesian method, and the corresponding model is dynamically selected according to the model matching probability to estimate the road adhesion coefficient, which specifically includes:

[0095] For the nonlinear system state estimation model composed of state variables x(t) and observation variables z(t) with Gaussian white noise w(t) and v(t), the road adhesion coefficient state equation and measurement equation are established as follows:

[0096]

[0097] Where f() and h() are the system state function and measurement function respectively; w(t) is the process noise at time t; v(t) is the measurement noise at time t; x(t) is the state variable at time t; z(t) is the observed variable at time t;

[0098] The road adhesion coefficient estimation algorithm based on Kalman filter takes the current road adhesion coefficient μ of each wheel as i As state variable x = (μ1, μ2, μ3, μ4) T ; For different driving conditions, different sensor signals are selected as observed variables. For straight-line driving conditions, the longitudinal acceleration a of the vehicle is selected.x and the angular acceleration of the wheel ω i As the observed variable z = (a x ,ω1,ω2,ω3,ω4) T ; For cornering conditions, select the vehicle's longitudinal acceleration a x , lateral acceleration a y and yaw rate γ as observed variables z = (a x ,a y ,γ) T , the sampling time of the Kalman filter is set to Δt, and the state space equation established after the system is discretized is as follows:

[0099]

[0100] In the formula, x(k+1) represents the state variable at time k+1; x(k) represents the state variable at time k; w(k) is the process noise at time k;

[0101] For the estimation of the road adhesion coefficient under different driving conditions, the conditions are divided into longitudinal conditions and transverse and longitudinal coupled conditions according to the driving state of the vehicle. The Bayesian method is used for model matching, the model matching probability under different conditions is calculated, and the corresponding model is dynamically selected for road adhesion coefficient estimation based on the model matching probability.

[0102] When the Bayesian method is used to calculate the model matching probability under different working conditions, the lateral acceleration and yaw rate are first taken as input features, the prior probability is set according to the vehicle's driving condition, and the lateral acceleration and yaw rate are set to obey the Gaussian distribution in each state, and the likelihood function is established. Then, the posterior probability in each state is calculated based on the Bayesian theorem, and the corresponding model is dynamically selected to estimate the road adhesion coefficient based on the maximum a posteriori probability criterion.

[0103] The longitudinal working condition model is used to estimate the longitudinal working condition of the vehicle. The lateral force is ignored in the observation equation, and the angular acceleration ω of the four wheels is introduced. i , set the observed variable of the longitudinal operating model to z x =(a x ,ω1,ω2,ω3,ω4) T , the observation equation of the longitudinal working condition model is:

[0104]

[0105] In the formula, z x (k) is the observed variable of the longitudinal operating model at time k; F xi 0 is the normalized tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; Fyi 0 is the normalized tire lateral force, where i = 1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; δ is the front wheel turning angle; m is the vehicle mass; R e is the wheel radius; I ω is the moment of inertia of the wheel around the axle; x(k) is the state variable at time k; T di is the driving torque; I ω is the moment of inertia of the wheel around the axle; v(k) is the measurement noise at time k; ψ is the longitudinal working condition model parameter, and ψ=(0,1,1,1,1) T ; cos is the cosine function.

[0106] The lateral and longitudinal coupling working condition model is used to estimate the longitudinal working condition of the vehicle. The longitudinal acceleration, lateral acceleration and yaw rate derivatives of the vehicle are selected as observation variables in the observation equation to obtain z xy =(a x ,a y ,γ) T ;

[0107] The observation equation of the transverse and longitudinal coupling model is:

[0108]

[0109] In the above formula, the parameters of the observation matrix are:

[0110]

[0111] In the formula, z xy (k) is the observed variable of the transverse and longitudinal coupling model at time k; F xi 0 is the normalized tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi 0 is the normalized tire lateral force, where i = 1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; δ is the front wheel turning angle; m is the vehicle mass; d f ,d r I is the front and rear wheelbase; z is the moment of inertia of the vehicle around the z-axis; a and b are the distances from the center of mass to the front and rear axes, respectively; H(3,1), H(3,2), H(3,3), H(3,4) are the observation matrix parameters; sin is the sine function; cos is the cosine function; d f ,d r Front and rear wheel tracks.

[0112] The adaptive square root volumetric Kalman filter algorithm is introduced to estimate the road adhesion coefficient, and the square root volumetric Kalman filter algorithm is constructed, including state prediction and measurement update, and an adaptive strategy based on the observed residual square sum and sliding window method is designed to calculate the residual and update the noise covariance matrix. The measurement noise covariance matrix is ​​updated in real time, and the process noise covariance matrix is ​​kept unchanged, specifically including:

[0113] According to the optimal estimate of the filter k-1 at the previous moment And the square root of the corresponding error covariance matrix S(k-1|k-1), the state estimate at time k is solved through time update and measurement update and the square root of the error covariance matrix S(k-1|k-1);

[0114] Combined with the sliding window method, the objective function based on the observation error is obtained as the iterative judgment condition. The corresponding calculation formula is as follows:

[0115]

[0116] Where J is the objective function based on observation error; n is the total number of sliding windows; W is the size of the sliding window; e(k) is the observation error at time k; e T (k) is the matrix transpose of the observation error at time k; η is the output value error weight matrix;

[0117] The calculation formula of the observation error e(k) at time k is as follows:

[0118]

[0119] Where, e(k) is the observation error at time k; z(k) is the observation variable at time k; is the optimal estimate of the observed variables from the previous time k-1 to time k;

[0120] Combined with the order of magnitude differences of different output values, the output value error weight matrix η is introduced:

[0121] In cornering conditions,

[0122]

[0123] Where η1, η2, η3 are the error weights of the output values;

[0124] In straight driving condition,

[0125]

[0126] Where η1 and η2 are the error weights of the output values;

[0127] By updating the measurement noise covariance matrix in real time and keeping the process noise covariance matrix unchanged, the measurement noise is updated as follows by combining the filtering process of the adaptive square root cubature Kalman filter algorithm and the adaptive strategy based on the sum of squares of the observed residuals and the sliding window method:

[0128]

[0129] Where H(k) is the residual of the observation at time k; e(k) is the observation error at time k; diag(e(k)) is the diagonal matrix of the observation error at time k; W is the size of the sliding window; is the adaptive smoothing factor; R(k) is the measurement noise covariance matrix at time k; R(k-1) is the measurement noise covariance matrix at the previous time k-1.

[0130] See also Figure 2 , is a road adhesion coefficient estimation system for an electric vehicle provided by an embodiment of the present invention, the system comprising:

[0131] A construction module is used to construct a 7-DOF model of a distributed drive electric vehicle based on the vehicle kinematic equations and wheel rotation equations, including longitudinal, lateral, yaw and rotational motion of the four wheels;

[0132] A fitting module is used to determine the nonlinear relationship between the vehicle motion state, the road adhesion coefficient and the tire force based on the Dugoff tire model, and obtain the normalized tire longitudinal and lateral forces by fitting the vehicle motion state;

[0133] An estimation module is used to determine the current working condition according to the driving state of the vehicle, construct a longitudinal working condition model and a transverse and longitudinal coupling working condition model for the straight working condition and the turning working condition respectively, establish the corresponding state equation and observation equation, and calculate the model matching probability under different working conditions based on the Bayesian method, and dynamically select the corresponding model according to the model matching probability to estimate the road adhesion coefficient;

[0134] The update module is used to introduce the adaptive square root volumetric Kalman filter algorithm to estimate the road adhesion coefficient, construct the square root volumetric Kalman filter algorithm, including state prediction and measurement update, design an adaptive strategy based on the sum of squares of observed residuals and the sliding window method, calculate the residuals and update the noise covariance matrix, update the measurement noise covariance matrix in real time, and keep the process noise covariance matrix unchanged;

[0135] The fusion module is used to fuse the estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model through a weighted average method to obtain a fused estimated value of the road adhesion coefficient.

[0136] Figure 2 The apparatus of the embodiment shown can be used to perform Figure 1 The implementation principles and technical effects of the steps in the method embodiment shown are similar and will not be repeated here.

[0137] See also Figure 3 , is a schematic diagram of the hardware structure of an electronic device provided in an embodiment of the present invention, the electronic device 30 includes: a processor 31, a memory 32 and a computer program; wherein

[0138] The memory 32 is used to store the computer program, and the memory may also be a flash memory. The computer program is, for example, an application program, a functional module, etc. for implementing the above method.

[0139] The processor 31 is used to execute the computer program stored in the memory to implement each step performed by the device in the above method. For details, please refer to the relevant description in the above method embodiment.

[0140] Optionally, the memory 32 may be independent or integrated with the processor 31 .

[0141] When the memory 32 is a device independent of the processor 31, the device may further include:

[0142] The bus 33 is used to connect the memory 32 and the processor 31 .

[0143] The present invention also provides a readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, it is used to implement the methods provided by the various embodiments described above.

[0144] Among them, the readable storage medium can be a computer storage medium or a communication medium. The communication medium includes any medium that facilitates the transmission of a computer program from one place to another. The computer storage medium can be any available medium that can be accessed by a general-purpose or special-purpose computer. For example, a readable storage medium is coupled to a processor so that the processor can read information from the readable storage medium and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can be located in an application-specific integrated circuit (Application Specific Integrated Circuits, referred to as: ASIC). In addition, the ASIC can be located in a user device. Of course, the processor and the readable storage medium can also exist in a communication device as discrete components. The readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, and an optical data storage device, etc.

[0145] The present invention also provides a program product, which includes an execution instruction, which is stored in a readable storage medium. At least one processor of a device can read the execution instruction from the readable storage medium, and at least one processor executes the execution instruction so that the device implements the methods provided in the above various embodiments.

[0146] In the embodiments of the above-mentioned devices, it should be understood that the processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), etc. The general-purpose processor may be a microprocessor or any conventional processor. The steps of the method disclosed in the present invention may be directly implemented as being executed by a hardware processor, or may be executed by a combination of hardware and software modules in the processor.

[0147] Through the introduction of the above embodiments, the present invention uses an electric vehicle road adhesion coefficient estimation method and system, and constructs a 7-degree-of-freedom model of a distributed drive electric vehicle based on the vehicle kinematic equation and the wheel rotation equation, including longitudinal, lateral, yaw and rotational motion of four wheels; determines the nonlinear relationship between the vehicle motion state, road adhesion coefficient and tire force based on the Dugoff tire model, and obtains the normalized tire longitudinal and lateral forces through vehicle motion state fitting; judges the current working condition according to the vehicle driving state, constructs a longitudinal working condition model and a transverse and longitudinal coupling working condition model for straight-line working conditions and turning working conditions, establishes corresponding state equations and observation equations, and calculates the model matching probability under different working conditions based on the Bayesian method, and dynamically selects the corresponding model to estimate the road adhesion coefficient according to the model matching probability; introduces an adaptive square root volumetric Kalman filter algorithm to estimate the road adhesion coefficient A square root volumetric Kalman filter algorithm is constructed, including state prediction and measurement update, an adaptive strategy based on the residual square sum of observations and the sliding window method is designed, the residual and the noise covariance matrix are calculated, the measurement noise covariance matrix is ​​updated in real time, and the process noise covariance matrix is ​​kept unchanged; the estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model are fused by the weighted average method to obtain the fused road adhesion coefficient estimation value, so that dynamic switching can be performed according to different driving conditions, thereby improving the adaptability and accuracy of the estimation; and by introducing an adaptive strategy and a Bayesian classification method, the stability and accuracy of the estimation are further improved, providing support for the active safety control of the vehicle; at the same time, the method of the present invention has broad application prospects, and can be applied to the fields of safety control systems, intelligent driving assistance systems, etc. of electric vehicles, to provide guarantee for the safe driving of electric vehicles.

[0148] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for estimating the road adhesion coefficient of an electric vehicle, characterized in that: The method comprises: Based on the vehicle kinematic equations and wheel rotation equations, a 7-DOF model of distributed drive electric vehicles is constructed, including longitudinal, lateral, yaw and rotational motion of the four wheels; Based on the Dugoff tire model, the nonlinear relationship between the vehicle motion state, road adhesion coefficient and tire force is determined, and the normalized tire longitudinal and lateral forces are obtained by fitting the vehicle motion state. The current working condition is determined according to the driving state of the vehicle, and a longitudinal working condition model and a transverse and longitudinal coupling working condition model are constructed for the straight working condition and the turning working condition respectively, and the corresponding state equation and observation equation are established. The model matching probability under different working conditions is calculated based on the Bayesian method, and the corresponding model is dynamically selected according to the model matching probability to estimate the road adhesion coefficient; Introduce an adaptive square root cubature Kalman filter algorithm to estimate the road adhesion coefficient, construct a square root cubature Kalman filter algorithm, including state prediction and measurement update, design an adaptive strategy based on the sum of squares of observation residuals and the sliding window method, calculate the residuals and update the noise covariance matrix, update the measurement noise covariance matrix in real time, and keep the process noise covariance matrix unchanged; The estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model are fused through the weighted average method to obtain the fused estimated value of the road adhesion coefficient.

2. The method for estimating the road adhesion coefficient of an electric vehicle according to claim 1, characterized in that: Based on the vehicle kinematic equation and wheel rotation equation, a 7-DOF model of a distributed drive electric vehicle is constructed, including longitudinal, lateral, yaw and rotational motion of four wheels, specifically including: The 7-DOF model of the distributed drive electric vehicle includes longitudinal, lateral, yaw and rotational motion of the four wheels; The vertical and lateral motions of the vehicle are ignored, and the vehicle is traveling on a horizontal road; the influence of the suspension on the vehicle is ignored, and the entire vehicle body is regarded as a rigid body; the influence of air resistance is ignored; the influence of the steering system is ignored, and the steering wheel angle is directly converted into the front wheel angle input. At this time, the vehicle kinematic equation is: For the rotational motion of a single wheel, according to the force on the wheel, the wheel rotation equation is: I ω ω i =T di -F xi R e (4) In the formula, v x is the longitudinal speed of the vehicle; v y is the lateral speed of the vehicle; a x is the longitudinal acceleration of the vehicle; a y is the lateral acceleration of the vehicle; γ is the yaw rate; m is the vehicle mass; F xi is the tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi is the tire lateral force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; a and b are the distances from the center of mass to the front and rear axles respectively; ω i is the wheel speed, i.e. the angular acceleration of the wheel; δ is the front wheel turning angle; d f d r I is the front and rear wheelbase; z is the moment of inertia of the vehicle around the z axis; I ω is the moment of inertia of the wheel around the axle; T di is the driving torque; R e is the wheel radius; sin is the sine function; cos is the cosine function.

3. The method for estimating the road adhesion coefficient of an electric vehicle according to claim 1, characterized in that: The method of determining the nonlinear relationship between the vehicle motion state, the road adhesion coefficient and the tire force based on the Dugoff tire model, and obtaining the normalized tire longitudinal and lateral forces by fitting the vehicle motion state, specifically includes: Based on the Dugoff tire model, the nonlinear relationship between the vehicle motion state, road adhesion coefficient and tire force is determined, and the road adhesion coefficient variable is separated. The normalized tire longitudinal and lateral forces are obtained by fitting the vehicle motion state, that is: In the formula, F x 、F y F are the longitudinal force and lateral force of the tire respectively; x 0 、F y 0 are the normalized tire longitudinal force and lateral force respectively; C x , C y are the longitudinal stiffness and lateral stiffness of the tire respectively; P(L) is the function of tire force used to correct the nonlinear interval; L is the tire working state parameter, L<1 means the tire is in the nonlinear region, otherwise it is in the linear region; κ and α are the slip rate and sideslip angle respectively; ε is the speed influence factor; μ is the road adhesion coefficient; F z is the vertical load of the tire; v x is the longitudinal speed of the vehicle; tan is the tangent function.

4. The method for estimating the road adhesion coefficient of an electric vehicle according to claim 1, characterized in that: The current working condition is judged according to the driving state of the vehicle, a longitudinal working condition model and a transverse and longitudinal coupling working condition model are constructed for the straight working condition and the turning working condition respectively, the corresponding state equation and observation equation are established, and the model matching probability under different working conditions is calculated based on the Bayesian method, and the corresponding model is dynamically selected according to the model matching probability to estimate the road adhesion coefficient, which specifically includes: For the nonlinear system state estimation model composed of state variables x(t) and observation variables z(t) with Gaussian white noise w(t) and v(t), the road adhesion coefficient state equation and measurement equation are established as follows: Where f() and h() are the system state function and measurement function respectively; w(t) is the process noise at time t; v(t) is the measurement noise at time t; x(t) is the state variable at time t; z(t) is the observed variable at time t; The road adhesion coefficient estimation algorithm based on Kalman filter takes the current road adhesion coefficient μ of each wheel as i As state variable x = (μ1, μ2, μ3, μ4) T ; For different driving conditions, different sensor signals are selected as observed variables. For straight-line driving conditions, the longitudinal acceleration a of the vehicle is selected. x and the angular acceleration of the wheel ω i As the observed variable z = (a x ,ω1,ω2,ω3,ω4) T ; For cornering conditions, select the vehicle's longitudinal acceleration a x , lateral acceleration a y and yaw rate γ as observed variables z = (a x ,a y ,γ) T , the sampling time of the Kalman filter is set to Δt, and the state space equation established after the system is discretized is as follows: In the formula, x(k+1) represents the state variable at time k+1; x(k) represents the state variable at time k; w(k) is the process noise at time k; For the estimation of the road adhesion coefficient under different driving conditions, the conditions are divided into longitudinal conditions and transverse and longitudinal coupled conditions according to the driving state of the vehicle. The Bayesian method is used for model matching, the model matching probability under different conditions is calculated, and the corresponding model is dynamically selected for road adhesion coefficient estimation based on the model matching probability.

5. The method for estimating the road adhesion coefficient of an electric vehicle according to claim 4, characterized in that: When the Bayesian method is used to calculate the model matching probability under different working conditions, the lateral acceleration and yaw rate are first taken as input features, the prior probability is set according to the vehicle's driving condition, and the lateral acceleration and yaw rate are set to obey the Gaussian distribution in each state, and the likelihood function is established. Then, the posterior probability in each state is calculated based on the Bayesian theorem, and the corresponding model is dynamically selected to estimate the road adhesion coefficient based on the maximum a posteriori probability criterion.

6. The method for estimating the road adhesion coefficient of an electric vehicle according to claim 4, characterized in that: The longitudinal working condition model is used to estimate the longitudinal working condition of the vehicle. The lateral force is ignored in the observation equation, and the angular acceleration ω of the four wheels is introduced. i , set the observed variable of the longitudinal operating model to z x =(a x ,ω1,ω2,ω3,ω4) T , the observation equation of the longitudinal working condition model is: In the formula, z x (k) is the observed variable of the longitudinal operating model at time k; F xi 0 is the normalized tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi 0 is the normalized tire lateral force, where i = 1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; δ is the front wheel turning angle; m is the vehicle mass; R e is the wheel radius; I ω is the moment of inertia of the wheel around the axle; x(k) is the state variable at time k; T di is the driving torque; I ω is the moment of inertia of the wheel around the axle; v(k) is the measurement noise at time k; ψ is the longitudinal working condition model parameter, and ψ=(0,1,1,1,1) T ; cos is the cosine function.

7. The method for estimating the road adhesion coefficient of an electric vehicle according to claim 4, characterized in that: The lateral and longitudinal coupling working condition model is used to estimate the longitudinal working condition of the vehicle. The longitudinal acceleration, lateral acceleration and yaw rate derivatives of the vehicle are selected as observation variables in the observation equation to obtain z xy =(a x ,a y ,γ) T ; The observation equation of the transverse and longitudinal coupling model is: In the above formula, the parameters of the observation matrix are: In the formula, z xy (k) is the observed variable of the transverse and longitudinal coupling model at time k; F xi 0 is the normalized tire longitudinal force, where i=1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; F yi 0 is the normalized tire lateral force, where i = 1, 2, 3, 4 represent the left front, right front, left rear, and right rear wheels respectively; δ is the front wheel turning angle; m is the vehicle mass; d f d r I is the front and rear wheelbase; z is the moment of inertia of the vehicle around the z-axis; a and b are the distances from the center of mass to the front and rear axes, respectively; H(3,1), H(3,2), H(3,3), H(3,4) are the observation matrix parameters; sin is the sine function; cos is the cosine function; d f d r Front and rear wheel tracks.

8. The method for estimating the road adhesion coefficient of an electric vehicle according to claim 1, characterized in that: The adaptive square root volumetric Kalman filter algorithm is introduced to estimate the road adhesion coefficient, and the square root volumetric Kalman filter algorithm is constructed, including state prediction and measurement update, and an adaptive strategy based on the residual square sum of observations and the sliding window method is designed to calculate the residual and update the noise covariance matrix. The measurement noise covariance matrix is ​​updated in real time, and the process noise covariance matrix is ​​kept unchanged, specifically including: According to the optimal estimate of the filter k-1 at the previous moment And the square root of the corresponding error covariance matrix S(k-1|k-1), the state estimate at time k is solved through time update and measurement update and the square root of the error covariance matrix S(k-1|k-1); Combined with the sliding window method, the objective function based on the observation error is obtained as the iterative judgment condition. The corresponding calculation formula is as follows: Where J is the objective function based on observation error; n is the total number of sliding windows; W is the size of the sliding window; e(k) is the observation error at time k; e T (k) is the matrix transpose of the observation error at time k; η is the output value error weight matrix; The calculation formula of the observation error e(k) at time k is as follows: Where, e(k) is the observation error at time k; z(k) is the observation variable at time k; is the optimal estimate of the observed variables from the previous time k-1 to time k; Combined with the order of magnitude differences of different output values, the output value error weight matrix η is introduced: In cornering conditions, Where η1, η2, η3 are the error weights of the output values; In straight driving condition, Where η1 and η2 are the error weights of the output values; By updating the measurement noise covariance matrix in real time and keeping the process noise covariance matrix unchanged, the measurement noise is updated as follows by combining the filtering process of the adaptive square root cubature Kalman filter algorithm and the adaptive strategy based on the sum of squares of the observed residuals and the sliding window method: Where H(k) is the residual of the observation at time k; e(k) is the observation error at time k; diag(e(k)) is the diagonal matrix of the observation error at time k; W is the size of the sliding window; is the adaptive smoothing factor; R(k) is the measurement noise covariance matrix at time k; R(k-1) is the measurement noise covariance matrix at the previous time k-1.

9. An estimation system for an electric vehicle road adhesion coefficient estimation method according to any one of claims 1 to 8, characterized in that: The system comprises: A construction module is used to construct a 7-DOF model of a distributed drive electric vehicle based on the vehicle kinematic equations and wheel rotation equations, including longitudinal, lateral, yaw and rotational motion of the four wheels; A fitting module is used to determine the nonlinear relationship between the vehicle motion state, the road adhesion coefficient and the tire force based on the Dugoff tire model, and obtain the normalized tire longitudinal and lateral forces by fitting the vehicle motion state; An estimation module is used to determine the current working condition according to the driving state of the vehicle, construct a longitudinal working condition model and a transverse and longitudinal coupling working condition model for the straight working condition and the turning working condition respectively, establish the corresponding state equation and observation equation, and calculate the model matching probability under different working conditions based on the Bayesian method, and dynamically select the corresponding model according to the model matching probability to estimate the road adhesion coefficient; The update module is used to introduce the adaptive square root volumetric Kalman filter algorithm to estimate the road adhesion coefficient, construct the square root volumetric Kalman filter algorithm, including state prediction and measurement update, design an adaptive strategy based on the sum of squares of observed residuals and the sliding window method, calculate the residuals and update the noise covariance matrix, update the measurement noise covariance matrix in real time, and keep the process noise covariance matrix unchanged; The fusion module is used to fuse the estimation results based on the longitudinal working condition model and the transverse and longitudinal coupling working condition model through a weighted average method to obtain a fused estimated value of the road adhesion coefficient.

Images