Vehicle adaptive model prediction control method

Abstract

The invention relates to a vehicle adaptive model prediction control method. The method comprises the following steps: collecting vehicle dynamic parameters, and constructing a two-degree-of-freedom vehicle dynamic model according to the dynamic parameters; constructing a first matrix used for describing an input and output relation of the linear state space equation; determining a model order of a vehicle model corresponding to the two-degree-of-freedom vehicle dynamics model through an akaike information criterion, and identifying parameters of the first matrix through a recursive least square method of introducing a forgetting factor to obtain the vehicle model; self-adaptively adjusting forgetting factors in the vehicle model through fuzzy logic and identification residual errors to obtain an optimized vehicle model; and establishing a vehicle controller based on the optimized vehicle model to control a vehicle control system in real time. According to the method, effective tracking of the expected state of the vehicle under different working conditions can be realized, the identification speed and the modeling precision are improved, and the modeling complexity is reduced.

Description

Technical Field

[0001] The invention belongs to the field of vehicle control, and in particular relates to a vehicle adaptive model predictive control method. Background Art

[0002] The design of vehicle control systems is closely related to the accuracy of the model, and the dynamics model is of great significance for realizing autonomous driving based on model predictive control.

[0003] Traditional vehicle dynamics models have been simplified in many ways to facilitate controller design. The entire vehicle is often simplified into a single vehicle model and only the linear characteristics of the tires are considered, which introduces a large number of unmodeled dynamics. Dynamic models with high degrees of freedom and considering the nonlinear characteristics of tires face problems such as high modeling complexity, poor versatility, and strong nonlinear coupling. Therefore, the choice of model requires a trade-off between modeling accuracy and controller design.

[0004] At the same time, traditional vehicle dynamics models cannot express road information, such as road slope and road friction coefficient, so they are all based on certain assumptions, such as assuming that the road slope change rate is zero, assuming that the longitudinal speed is constant, assuming that the sideslip angle is in the linear working area, etc., which has certain limitations. When the working conditions do not meet the modeling assumptions, model mismatch will occur. At the same time, the modeling difficulty brought by traditional vehicle dynamics models cannot be ignored. There are many vehicle parameters in actual vehicles that cannot be directly obtained, such as axle lateral stiffness, roll stiffness, roll damping, center of mass position, moment of inertia and other vehicle parameters. It is very complex and difficult to accurately model the dynamics models of different vehicles.

[0005] During the actual driving process of the vehicle, changes in factors such as the external environment (such as the road friction coefficient), vehicle speed, and driving conditions will affect the vehicle's driving state, causing some of the vehicle's structural parameters (such as tire stiffness and mass) to change and become uncertain. Under extreme conditions, the uncertainty of the vehicle's structural parameters will be even stronger, resulting in errors between vehicle modeling and actual engineering applications, causing model mismatch. The original vehicle model cannot represent the current state. Using a certain vehicle structural parameter value will make it difficult for the control strategy designed based on the vehicle model to match the actual driving conditions, and the control effect will not meet the expected requirements, affecting the precise control of the vehicle. Summary of the invention

[0006] In view of the above deficiencies in the prior art, an object of the invention is to provide a vehicle adaptive model predictive control method.

[0007] The present invention provides a vehicle adaptive model predictive control method, comprising:

[0008] S1: collecting vehicle dynamics parameters, and constructing a two-degree-of-freedom vehicle dynamics model according to the dynamics parameters;

[0009] S2: constructing a first matrix for describing the input-output relationship of the linear state space equation, wherein the dimension of the first matrix is ​​equal to the number of inputs and outputs of the two-degree-of-freedom vehicle dynamics model;

[0010] S3: determining the model order of the vehicle model corresponding to the two-degree-of-freedom vehicle dynamics model by Akaike information criterion, identifying the parameters of the first matrix by a recursive least squares method with a forgetting factor introduced, and obtaining the vehicle model;

[0011] S4: adaptively adjusting the forgetting factor in the vehicle model through fuzzy logic and identification residual to obtain an optimized vehicle model;

[0012] S5: Establishing a vehicle controller based on the optimized vehicle model, and controlling the vehicle control system in real time through the vehicle controller.

[0013] According to a vehicle adaptive model predictive control method provided by the present invention, the dynamic parameters in step S1 include a steering wheel angle input, a vehicle lateral velocity output, and a yaw angular velocity output sequence.

[0014] According to a vehicle adaptive model predictive control method provided by the present invention, the first matrix in step S2 is a discrete time z transfer function matrix.

[0015] According to a vehicle adaptive model predictive control method provided by the present invention, the expression of the first matrix in step S2 is:

[0016]

[0017] Among them, G(z) is the first matrix related to the state-space equation, z is the z-transformation factor, I is the identity matrix, (A, B, C, D) represents the system matrix, Q(z) is the numerator polynomial matrix, and α(z) is the characteristic polynomial representing the least common denominator of the first matrix.

[0018] According to a vehicle adaptive model predictive control method provided by the present invention, the expression of the Akaike information criterion in step S3 is:

[0019] AIC = 2k-2ln(L(θ));

[0020] Among them, AIC is the Akaike information criterion, k is the total number of parameters that need to be estimated in the model, θ is the parameter to be estimated, and ln(L(θ)) is the log-likelihood function value of the model.

[0021] According to a vehicle adaptive model predictive control method provided by the present invention, step S3 further comprises:

[0022] S31: constructing an identification system equation according to the model order, wherein the expression of the identification system equation is:

[0023]

[0024] Among them, y(k) is the output vector of the identification system equation, H(k) is the input and output data information matrix at time k, θ s is the parameter vector to be identified at time k, ψ(k) is the output information matrix, is the transposed matrix of the input information vector, I is the identity matrix, α is the output parameter vector, θ T is the transposed matrix of the input parameter matrix, vec(θ T ) is θ T Perform Kronecker product transformation to obtain the stacked vector;

[0025] S32: Performing offline parameter analysis according to the identification system equation to obtain an offline identification equation, wherein the expression of the offline identification equation is:

[0026]

[0027] in, is the unique solution obtained by the least squares parameter estimation corresponding to the offline identification equation, N is the total number of model orders, Φ N is the matrix of input and output data information of all orders, is the transposed matrix of the matrix of input and output data information of all orders, y N is the output vector matrix of all orders;

[0028] S33: Perform online parameter analysis according to the identification system equation and the offline identification equation, and introduce a forgetting factor to obtain an online identification equation. The expression of the online identification equation is:

[0029]

[0030] in, is the unique solution obtained by the least squares parameter estimation corresponding to the online identification equation at time k, is the unique solution obtained by the least squares parameter estimation corresponding to the online identification equation at time k-1, K(k) is the Kalman gain matrix at time k, P(k-1) is the covariance matrix at time k-1, H T (k) is the transposed matrix of the input and output data information matrix at time k, λ is the forgetting factor, H T(k-1) is the transposed matrix of the input and output data information matrix at time k-1, and P(k) is the covariance matrix at time k;

[0031] S34: Outputting the online identification equation as the vehicle model.

[0032] According to a vehicle adaptive model predictive control method provided by the present invention, step S4 further comprises:

[0033] S41: using the output residual of the yaw angular velocity in the identified vehicle model and the vehicle lateral velocity as input values, and using the forgetting factor as an output value, to design a fuzzy controller;

[0034] S42: Adapting the forgetting factor in the vehicle model by the forgetting factor calculated by the fuzzy controller.

[0035] According to a vehicle adaptive model predictive control method provided by the present invention, the vehicle controller in step S5 is a double-layer controller, the upper layer of the double-layer controller is a lane keeping model, and the lower layer of the double-layer controller is a tracking control model.

[0036] According to a vehicle adaptive model predictive control method provided by the present invention, step S5 further comprises:

[0037] S51: Select the center line of the road as a reference line and establish a lane keeping model;

[0038] S52: Outputting a desired state quantity sequence through the lane keeping model;

[0039] S53: inputting the desired state quantity sequence into the tracking control model to obtain the vehicle front wheel angle control quantity;

[0040] S54: Inputting the vehicle front wheel steering angle control amount into the vehicle control system to perform real-time control on the vehicle.

[0041] According to a vehicle adaptive model predictive control method provided by the present invention, the expression of the lane keeping model in step S51 is:

[0042]

[0043] Among them, e y (k+1) is the expected state of the vehicle lateral velocity at time k+1, e y (k) is the expected state of the vehicle lateral velocity at time k, Ts is the sampling time, v x is the longitudinal velocity of the vehicle, e φ (k) is the expected state of the vehicle yaw rate at time k, v y(k) is the lateral speed of the vehicle at time k, L is the preview distance, e φ (k+1) is the expected state of the vehicle yaw rate at time k+1, is the desired vehicle yaw rate, is the actual vehicle yaw rate.

[0044] The beneficial effects of the present invention are as follows:

[0045] The present invention provides a vehicle adaptive model predictive control method, which avoids unmodeled dynamics by directly identifying the mapping relationship between input and output data online. At the same time, by continuously iterating online to identify this mapping relationship, it can explicitly consider the impact of changes in the external environment and the overall structural parameters of the vehicle body, thereby suppressing parameter uncertainty to a certain extent. By establishing an accurate linear vehicle dynamics model at each working point, the vehicle's adaptive model predictive control is realized, thereby improving control accuracy while ensuring computational efficiency.

[0046] In addition, the modeling method provided by the present invention has lower modeling complexity and modeling difficulty, and does not need to be established based on assumptions. It only needs to collect specific input and output data of different vehicles under different working conditions to establish a dynamic model suitable for different vehicles. Therefore, it can be easily transplanted to other vehicles to complete the modeling work, and has strong generalization and universality.

[0047] The present invention is different from Jacobi linearization. The accuracy of Jacobi linearization still depends on the accuracy of the original nonlinear model. If the original nonlinear model has unpredictable unmodeled dynamics at this working point, the model after Jacobi linearization will not be able to accurately characterize the current state, thereby affecting the precise control of the vehicle. The method described in the present invention is a data-driven working point linear modeling method, which establishes a linear model at each working point based on data at the current moment and the past moment. While having the original dynamic characteristics, it can characterize the dynamic characteristics of the current working condition to a certain extent, and is robust to changes in the external environment and the overall structural parameters of the vehicle. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] The accompanying drawings are only used to illustrate specific embodiments and are not considered to limit the present invention. In the entire drawings, the same reference symbols represent the same components. Obviously, the drawings described below are only some embodiments recorded in the embodiments of the present invention. For those of ordinary skill in the art, other drawings can also be obtained based on these drawings.

[0049] Figure 1 The present invention is a flow chart of a vehicle adaptive model predictive control method according to an embodiment of the present invention. DETAILED DESCRIPTION

[0050] In order to enable those skilled in the art to better understand the technical solutions in the embodiments of the present invention, the technical solutions of the present invention will be clearly and completely described below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all of the embodiments. It should be understood that these descriptions are only exemplary and are not intended to limit the scope of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in the field without making creative work should fall within the scope of protection of the present invention.

[0051] Furthermore, in the following description, descriptions of well-known structures and technologies are omitted to avoid unnecessarily obscuring the concepts disclosed in the present invention.

[0052] In the description of the present invention, it should be noted that, unless otherwise clearly specified and limited, the orientation or position relationship indicated by the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. is based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, and does not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation of the present invention. In addition, the terms "first", "second", and "third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. The terms "installed", "connected", and "connected" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be a connection between the two elements. For those of ordinary skill in the art, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0053] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of methods and systems consistent with some aspects of the present invention as detailed in the appended claims.

[0054] Combine the following Figure 1 Embodiments of the present invention are described.

[0055] like Figure 1 As shown, the present invention provides a vehicle adaptive model predictive control method, comprising:

[0056] S1: collecting vehicle dynamics parameters, and constructing a two-degree-of-freedom vehicle dynamics model according to the dynamics parameters;

[0057] The dynamic parameters in step S1 include a steering wheel angle input, a vehicle lateral velocity output, and a yaw rate output sequence.

[0058] S2: constructing a first matrix for describing the input-output relationship of the linear state space equation, wherein the dimension of the first matrix is ​​equal to the number of inputs and outputs of the two-degree-of-freedom vehicle dynamics model;

[0059] Wherein, the first matrix in step S2 is a discrete time z transfer function matrix.

[0060] The expression of the first matrix in step S2 is:

[0061]

[0062] Among them, G(z) is the first matrix related to the state-space equation, z is the z-transformation factor, I is the identity matrix, (A, B, C, D) represents the system matrix, Q(z) is the numerator polynomial matrix, and α(z) is the characteristic polynomial representing the least common denominator of the first matrix.

[0063] In step S2, a discrete time z-transfer function matrix model based on adaptive forgetting factor recursive least squares algorithm identification is first constructed, specifically, the following discrete time z-transfer function matrix equation describing the input / output relationship is first considered:

[0064] y(k)=G(z)u(k),

[0065] Among them, u(k) is the system input vector, y(k) is the system output vector, and G(z) is the discrete time z-transfer function matrix associated with the state space equation, that is, the first matrix mentioned above, which is in the following form:

[0066]

[0067] Where z is the z-transformation factor, α(z) is the characteristic polynomial, defined as the lowest common denominator of G(z), and Q(z) is the numerator polynomial matrix, both of which are expressed as follows:

[0068]

[0069] Among them, z -1 is the backshift operator, α i (i=1,…,n α ) and Q j (j=1,…,n Q ) are the unknown coefficients of the polynomials α(z) and Q(z), n α and n Q are the orders of the polynomials α(z) and Q(z), respectively, and we can get:

[0070] α(z)y(k)=Q(z)u(k),

[0071]

[0072] Subsequently, the model order can be determined offline using this formula and the AIC criterion.

[0073] S3: determining the model order of the vehicle model corresponding to the two-degree-of-freedom vehicle dynamics model by Akaike information criterion, identifying the parameters of the first matrix by a recursive least squares method with a forgetting factor introduced, and obtaining the vehicle model;

[0074] The expression of the Akaike information criterion in step S3 is:

[0075] AIC = 2k-2ln(L(θ));

[0076] Among them, AIC is the Akaike information criterion, k is the total number of parameters that need to be estimated in the model, θ is the parameter to be estimated, and ln(L(θ)) is the log-likelihood function value of the model.

[0077] In the above formula, k is the total number of parameters that need to be estimated in the model, representing the complexity of the model, ln(L(θ)) is the log-likelihood function value of the model, representing the model fitting ability, and the smaller the value obtained by AIC calculation, the better the balance between fitting data and complexity of the model.

[0078] At the initial moment, the steering wheel angle input, vehicle lateral velocity output and yaw rate output sequence of the driver when driving on the serpentine road are collected, and the model order n that best suits this type of data is identified offline using AIC. α and n Q , the present invention will n α and n Q The same value is determined, which is defined as the model order n. In a specific embodiment, the model order is determined to be 2.

[0079] Wherein, step S3 further comprises:

[0080] S31: constructing an identification system equation according to the model order, wherein the expression of the identification system equation is:

[0081]

[0082] Among them, y(k) is the output vector of the identification system equation, H(k) is the input and output data information matrix at time k, θ s is the parameter vector to be identified at time k, ψ(k) is the output information matrix, is the transposed matrix of the input information vector, I is the identity matrix, α is the output parameter vector, θ T is the transposed matrix of the input parameter matrix, vec(θ T ) is θ T Perform the Kronecker product transformation to obtain the stacked vector.

[0083] In step S31, first define the input parameter matrix θ, the output parameter vector α, and the input information vector Then the output information matrix ψ(k) can be obtained as follows:

[0084]

[0085] The system parameters to be identified include a parameter vector α composed of characteristic polynomial coefficients and a parameter matrix θ composed of numerator polynomial matrix coefficients. T Perform Kronecker product transformation into a stacked vector vec(θ T ), rewrite the above equation into the above identification system equation:

[0086]

[0087] Among them, I m is an m×m dimensional unit matrix, which combines the input and output data information matrix H(k) with the parameter vector θ to be identified s The product form of is used as the constructed identification system equation, and the identification system equation is determined to be a second-order system. Finally, the parameter θ s is regarded as the parameter to be identified in the identification system equation.

[0088] S32: Performing offline parameter analysis according to the identification system equation to obtain an offline identification equation, wherein the expression of the offline identification equation is:

[0089]

[0090] in, is the unique solution obtained by the least squares parameter estimation corresponding to the offline identification equation, N is the total number of model orders, Φ N is the matrix of input and output data information of all orders, is the transposed matrix of the matrix of input and output data information of all orders, y N is the output vector matrix of all orders.

[0091] In step S32, an offline parameter identification equation is analyzed according to the identification system equation to obtain a corresponding offline identification equation. The specific process is as follows:

[0092] First, construct the residual ∈(k,θs )for:

[0093]

[0094] Secondly, the error cost function is constructed as the sum of squares of the residuals:

[0095]

[0096] The least squares parameter estimates are:

[0097]

[0098] in,

[0099]

[0100] The error cost function can be rewritten as:

[0101]

[0102] Parameter estimation that minimizes the error cost function The following canonical equation should be satisfied:

[0103]

[0104] if Full rank, we get a unique solution:

[0105]

[0106] The expression The output as the offline parameter identification equation is the offline identification equation obtained in step S32.

[0107] S33: Perform online parameter analysis according to the identification system equation and the offline identification equation, and introduce a forgetting factor to obtain an online identification equation. The expression of the online identification equation is:

[0108]

[0109] in, is the unique solution obtained by the least squares parameter estimation corresponding to the online identification equation at time k, is the unique solution obtained by the least squares parameter estimation corresponding to the online identification equation at time k-1, K(k) is the Kalman gain matrix at time k, P(k-1) is the covariance matrix at time k-1, H T (k) is the transposed matrix of the input and output data information matrix at time k, λ is the forgetting factor, H T(k-1) is the transposed matrix of the input and output data information matrix at time k-1, and P(k) is the covariance matrix at time k.

[0110] S34: Outputting the online identification equation as the vehicle model.

[0111] In steps S33 to S34, online parameter identification equations are analyzed according to the identification system equations and offline parameter identification equations to obtain corresponding online identification equations. The specific process is to first determine the Kalman gain matrix and the covariance matrix, and then the obtained linear recursive equations and the corresponding Kalman gain matrix K(k) and covariance matrix P(k) and input and output data information matrices constitute the online recursive identification equation group.

[0112] S4: adaptively adjusting the forgetting factor in the vehicle model through fuzzy logic and identification residual to obtain an optimized vehicle model;

[0113] Wherein, step S4 further comprises:

[0114] S41: using the output residual of the yaw angular velocity in the identified vehicle model and the vehicle lateral velocity as input values, and using the forgetting factor as an output value, to design a fuzzy controller;

[0115] S42: Adapting the forgetting factor in the vehicle model by the forgetting factor calculated by the fuzzy controller.

[0116] The purpose of step S4 is to adaptively adjust the forgetting factor based on fuzzy logic and identification residuals to avoid algorithm oscillation convergence, covariance matrix saturation and "wind-up" problems caused by a fixed forgetting factor.

[0117] In step S41 to step S42, the identified vehicle lateral velocity and yaw rate output residuals are first used as the input of the fuzzy controller, and the value of the forgetting factor is used as the output, and a dual-input single-output fuzzy controller is designed to adaptively adjust and calculate the value of the forgetting factor. In the process of fuzzy rule generation, five fuzzy languages ​​{NB, NS, Z, PS, PB} are selected to describe the output residuals, and the fuzzy sets are NB (negative large), NS (negative small), Z (zero), PS (positive small), PB (positive large), and the input variables use Z-type membership function, π-type membership function and S-type membership function, and then three fuzzy languages ​​{PS, PM, PB} are selected to describe the value of the forgetting factor, and the fuzzy sets are PS (positive small), PM (positive middle), PB (positive large), and the output variables all use triangular membership functions. The defuzzification uses the area centroid method to defuzzify and refine the fuzzy values ​​generated by fuzzy reasoning, and finally obtain the accurate value of the forgetting factor.

[0118] S5: Establishing a vehicle controller based on the optimized vehicle model, and controlling the vehicle control system in real time through the vehicle controller.

[0119] Among them, the vehicle controller in step S5 is a double-layer controller, the upper layer of the double-layer controller is a lane keeping model, and the lower layer of the double-layer controller is a tracking control model.

[0120] Wherein, step S5 further comprises:

[0121] S51: Select the center line of the road as a reference line and establish a lane keeping model;

[0122] The lane keeping model in step S51 is expressed as:

[0123]

[0124] Among them, e y (k+1) is the expected state of the vehicle lateral velocity at time k+1, e y (k) is the expected state of the vehicle lateral velocity at time k, T s is the sampling time, v x is the longitudinal velocity of the vehicle, e φ (k) is the expected state of the vehicle yaw rate at time k, v y (k) is the lateral speed of the vehicle at time k, L is the preview distance, e φ (k+1) is the expected state of the vehicle yaw rate at time k+1, is the desired vehicle yaw rate, is the actual vehicle yaw rate.

[0125] In step S51, a lane keeping model of the vehicle relative to the center line of the road is first constructed, and then the discrete time z-transfer function matrix model is jointly augmented into a discrete vehicle linear time-varying model.

[0126] S52: Outputting a desired state quantity sequence through the lane keeping model.

[0127] The cost function of the upper layer desired state sequence planner of the lane keeping model in step S51 is:

[0128]

[0129] Among them, Q and R are weight matrices, N p is the prediction time domain, N c To control the time domain;

[0130]

[0131] Define the control quantity constraint: ur,min (k)≤u r (k)≤u r,max (k), then the expected state sequence U r,k Output to the underlying tracking controller.

[0132] S53: inputting the desired state quantity sequence into the tracking control model to obtain the vehicle front wheel angle control quantity;

[0133] In step S53, a lower-level tracking controller based on a discrete-time z-transfer function matrix model is constructed, and its cost function is:

[0134]

[0135] Among them, Q and R are weight matrices, N p is the prediction time domain, N c To control the time domain;

[0136]

[0137] u(k)=δ(k),

[0138] Define the control quantity constraint: u min (k)≤u(k)≤u max (k), that is, the output quantity constraint: y min (k)≤y(k)≤y max (k), and finally output the actual control amount δ of the front wheel steering angle to the actuator of the vehicle.

[0139] S54: Inputting the vehicle front wheel steering angle control amount into the vehicle control system to perform real-time control on the vehicle.

[0140] In a specific embodiment, the present invention conducts a specific lane keeping simulation experiment, in which the initial state of the vehicle is: y =0m / s,v x,min =11m / s,v x,min =25m / s, The road friction coefficient varies in the range of 0.85±0.3, the maximum turning radius is 80m, the preview distance is 8m, and the longitudinal speed of the vehicle in the experiment covers 40km / h~90km / h. It is expected that the vehicle will always travel near the center line of the road without exceeding the road boundary line. Based on the above parameters, the lane keeping simulation experiment is carried out. The above online identification method can effectively realize the accurate characterization of the vehicle state at different working points, and the above hierarchical control method can effectively realize the tracking control of the vehicle to the expected road center line.

[0141] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the embodiments of the present invention, rather than to limit them. Although the present invention has been described in detail with reference to the above embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the above embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention. Any changes or replacements that can be easily thought of by those skilled in the art within the technical scope disclosed in the present invention should be covered within the protection scope of the present invention.

Claims

1. A vehicle adaptive model predictive control method, characterized in that: include: S1: collecting vehicle dynamics parameters, and constructing a two-degree-of-freedom vehicle dynamics model according to the dynamics parameters; S2: constructing a first matrix for describing the input-output relationship of the linear state space equation, wherein the dimension of the first matrix is ​​equal to the number of inputs and outputs of the two-degree-of-freedom vehicle dynamics model; S3: determining the model order of the vehicle model corresponding to the two-degree-of-freedom vehicle dynamics model by Akaike information criterion, identifying the parameters of the first matrix by a recursive least squares method with a forgetting factor introduced, and obtaining the vehicle model; S4: adaptively adjusting the forgetting factor in the vehicle model through fuzzy logic and identification residual to obtain an optimized vehicle model; S5: Establishing a vehicle controller based on the optimized vehicle model, and controlling the vehicle control system in real time through the vehicle controller.

2. A vehicle adaptive model predictive control method according to claim 1, characterized in that: The dynamic parameters in step S1 include a steering wheel angle input, a vehicle lateral velocity output, and a yaw rate output sequence.

3. A vehicle adaptive model predictive control method according to claim 1, characterized in that: The first matrix in step S2 is a discrete time z transfer function matrix.

4. The vehicle adaptive model predictive control method according to claim 1, characterized in that: The expression of the first matrix in step S2 is: Among them, G(z) is the first matrix related to the state-space equation, z is the z-transformation factor, I is the identity matrix, (A, B, C, D) represents the system matrix, Q(z) is the numerator polynomial matrix, and α(z) is the characteristic polynomial representing the least common denominator of the first matrix.

5. The vehicle adaptive model predictive control method according to claim 1, characterized in that: The expression of the Akaike information criterion in step S3 is: AIC = 2k-2ln(L(θ)); Among them, AIC is the Akaike information criterion, k is the total number of parameters that need to be estimated in the model, θ is the parameter to be estimated, and ln(L(θ)) is the log-likelihood function value of the model.

6. A vehicle adaptive model predictive control method according to claim 1, characterized in that: Step S3 further comprises: S31: constructing an identification system equation according to the model order, wherein the expression of the identification system equation is: Among them, y(k) is the output vector of the identification system equation, H(k) is the input and output data information matrix at time k, θ s is the parameter vector to be identified at time k, ψ(k) is the output information matrix, is the transposed matrix of the input information vector, I is the identity matrix, α is the output parameter vector, θ T is the transposed matrix of the input parameter matrix, vec(θ T ) is θ T Perform Kronecker product transformation to obtain the stacked vector; S32: Performing offline parameter analysis according to the identification system equation to obtain an offline identification equation, wherein the expression of the offline identification equation is: in, is the unique solution obtained by the least squares parameter estimation corresponding to the offline identification equation, N is the total number of model orders, Φ N is the matrix of input and output data information of all orders, is the transposed matrix of the matrix of input and output data information of all orders, y N is the output vector matrix of all orders; S33: Perform online parameter analysis according to the identification system equation and the offline identification equation, and introduce a forgetting factor to obtain an online identification equation. The expression of the online identification equation is: in, is the unique solution obtained by the least squares parameter estimation corresponding to the online identification equation at time k, is the unique solution obtained by the least squares parameter estimation corresponding to the online identification equation at time k-1, K(k) is the Kalman gain matrix at time k, P(k-1) is the covariance matrix at time k-1, H T (k) is the transposed matrix of the input and output data information matrix at time k, λ is the forgetting factor, H T (k-1) is the transposed matrix of the input and output data information matrix at time k-1, and P(k) is the covariance matrix at time k; S34: Outputting the online identification equation as the vehicle model.

7. The vehicle adaptive model predictive control method according to claim 1, characterized in that: Step S4 further comprises: S41: using the output residual of the yaw angular velocity in the identified vehicle model and the vehicle lateral velocity as input values, and using the forgetting factor as an output value, to design a fuzzy controller; S42: Adapting the forgetting factor in the vehicle model by the forgetting factor calculated by the fuzzy controller.

8. The vehicle adaptive model predictive control method according to claim 1, characterized in that: The vehicle controller in step S5 is a double-layer controller, the upper layer of the double-layer controller is a lane keeping model, and the lower layer of the double-layer controller is a tracking control model.

9. The vehicle adaptive model predictive control method according to claim 7, characterized in that: Step S5 further comprises: S51: Select the center line of the road as a reference line and establish a lane keeping model; S52: Outputting a desired state quantity sequence through the lane keeping model; S53: inputting the desired state quantity sequence into the tracking control model to obtain the vehicle front wheel angle control quantity; S54: Inputting the vehicle front wheel steering angle control amount into the vehicle control system to perform real-time control on the vehicle.

10. A vehicle adaptive model predictive control method according to claim 9, characterized in that: The expression of the lane keeping model in step S51 is: Among them, e y (k+1) is the expected state of the vehicle lateral velocity at time k+1, e y (k) is the expected state of the vehicle lateral velocity at time k, Ts is the sampling time, v x is the longitudinal velocity of the vehicle, e φ (k) is the expected state of the vehicle yaw rate at time k, v y (k) is the lateral speed of the vehicle at time k, L is the preview distance, e φ (k+1) is the expected state of the vehicle yaw rate at time k+1, is the desired vehicle yaw rate, is the actual vehicle yaw rate.

Images