Intelligent automobile time-delay steering path tracking control method based on LPV model

Abstract

The invention discloses an intelligent automobile time-lag steering path tracking control method based on an LPV model. The method comprises the following steps: acquiring an extended state matrix considering steering delay characteristics; obtaining a path tracking error model based on an LPV module; and designing a model prediction controller objective function. Based on the monorail vehicle transverse dynamics model, the key parameter change of the vehicle in the path tracking process can be dynamically captured, so that more accurate system dynamic description is provided in the model prediction control process, the prediction precision and the accuracy of control response are remarkably improved, and the method is suitable for popularization and application. And the path tracking stability and the control precision of the intelligent automobile are ensured. According to the method, an extended state modeling and control increment constraint mechanism is introduced, effective modeling and feed-forward compensation of steering lag are achieved, the adaptability of a control system to the dynamic characteristics of an actual actuator is improved, the fault-tolerant capability of a controller to non-ideal execution lag behaviors is also enhanced, and the method is suitable for large-scale popularization and application. And the robustness and the response coordination in the path tracking process are ensured.

Description

TECHNICAL FIELD

[0001] The application relates to path tracking control technology of an intelligent automobile, in particular to an intelligent automobile time-delay steering path tracking control method based on an LPV model. BACKGROUND

[0002] An intelligent automobile control system is a core link of automatic driving, mainly undertakes the task of accurately converting the expected trajectory of a path planning layer into control instructions of each actuator, and the control precision and robustness of the intelligent automobile control system directly affect the safety and stability of the whole vehicle in a dynamic environment. In an actual path tracking process, the vehicle often faces various disturbance factors, such as dynamic changes of vehicle parameters (such as mass, cornering stiffness and yaw moment of inertia), external disturbances caused by a complex environment, response lag of an actuator, and delay in sensor measurement and signal transmission, which may all cause path tracking errors or even vehicle stability problems.

[0003] To cope with the above challenges, model predictive control (MPC) has been widely applied in vehicle path tracking control due to its good rolling optimization ability and input / state constraint processing ability. However, a traditional MPC algorithm is usually based on a fixed linear model to construct a system prediction model, and it is difficult to adapt to the time-varying characteristics of vehicle dynamic parameters, which limits its control performance in a strong nonlinear and parameter uncertain environment. In addition, the response delay phenomenon of a steering system as the main actuator of vehicle path control is particularly obvious under high speed or complex working conditions, and if the controller fails to model the time-delay characteristics, the prediction model will be deviated, thereby weakening the real-time performance and robustness of the system.

[0004] To improve the adaptability of the model predictive controller under the conditions of parameter perturbation and actuator lag, the linear parameter varying (LPV) modeling method has attracted widespread attention in recent years. The method constructs a parameter-dependent model by introducing time-varying parameters (such as cornering stiffness, mass and yaw moment of inertia), effectively describes the dynamic change characteristics of the vehicle system, and further improves the accuracy and robustness of the controller under different working conditions. Combined with the modeling method of a first-order lag system to process the steering delay, and the characteristic is embedded in the extended state space, the modeling accuracy of the system to the response dynamics of the actuator can be further improved. Therefore, it is of important engineering application value and theoretical significance to study a path tracking control method that fuses LPV modeling and MPC optimization structure and can consider the response time-delay characteristics of the steering actuator. SUMMARY

[0005] In order to solve the problems of the prior art, such as insufficient modeling of multi-channel disturbance, poor adaptability of vehicle parameter changes, and time delay of steering actuators, which have adverse effects on the performance of path tracking control, the present application aims to provide an intelligent vehicle time-delay steering path tracking control method based on an LPV model, which introduces a dynamic parameter scheduling mechanism and time delay modeling to accurately depict the changes in vehicle dynamic characteristics and the response delay of actuators, thereby improving the accuracy, stability and robustness of path tracking control.

[0006] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0007] An intelligent vehicle time-delay steering path tracking control method based on an LPV model, which utilizes an intelligent vehicle time-delay steering path tracking control system for control, the intelligent vehicle time-delay steering path tracking control system comprising an initialization module, an extended system model construction module, a linear parameter time-varying module, and a model predictive controller.

[0008] The initialization module is responsible for checking whether the signals of the perception module, mapping and positioning module, decision and planning module, and actuators are normal, loading single-track vehicle lateral dynamics model parameters, establishing a vehicle coordinate system, loading tracking error dynamics model parameters, and loading model predictive controller parameters.

[0009] The extended system model construction module constructs an extended state space model based on the tracking error dynamics model. On the basis of the original system state variables, a steering angle state variable is introduced, and a first-order lag system is used to model the actual front wheel steering angle, thereby forming an extended system model containing steering dynamics.

[0010] The linear parameter time-varying module dynamically generates a state space model required for predictive control according to the current state variables of the vehicle during vehicle operation.

[0011] The model predictive controller solves a quadratic programming problem based on the extended module and the LPV module to obtain the optimal front wheel steering angle control law.

[0012] The control method comprises the following steps:

[0013] Step 1: Initialization

[0014] The initialization module is responsible for checking whether the signal communication state of the perception module, the mapping and localization module, the decision and planning module, and the actuator is normal. The single-track vehicle lateral dynamics model parameters, the model predictive controller parameters, and the feedforward compensator parameters are loaded. The position information, the heading angle information, and the road curvature information of the path matching point sent by the decision and planning module are received. The path matching point is the projection point of the current intelligent vehicle position on the expected path, that is, the nearest point to the expected path.

[0015] The vehicle coordinate system is established as follows: taking the vehicle mass center as the origin, setting the x-axis positive direction to point to the vehicle forward direction, the y-axis positive direction to point to the vehicle left side direction, and the z-axis to be vertical upward, and the x-axis and the y-axis to jointly constitute a right-handed rectangular coordinate system.

[0016] The single-track vehicle lateral dynamics model is as follows:

[0017]

[0018] In the formula, C f and C r are the cornering stiffness of the front and rear wheels of the vehicle; m is the mass of the vehicle; v x is the longitudinal speed of the vehicle; y is the displacement in the lateral direction of the vehicle coordinate system, is the speed in the lateral direction of the vehicle coordinate system, is the acceleration in the lateral direction of the vehicle coordinate system; is the vehicle yaw angle, is the vehicle yaw rate, is the vehicle yaw angular acceleration; δ f is the front wheel steering angle of the vehicle; I z is the rotational inertia of the vehicle mass center around the Z-axis, and the Z-axis and the X-axis and the Y-axis in the vehicle coordinate system constitute a right-handed system; a and b are the distances from the vehicle mass center to the front and rear axles, respectively.

[0019] Based on the position information, the heading information, and the curvature at the expected path matching point, a tracking error dynamics model considering multi-channel uncertain disturbance is established as follows:

[0020]

[0021] In the formula, x(t) is the state variable in the vehicle path tracking control, x(t) is e d is the lateral error, is the first order derivative of the lateral error with respect to time, is the heading error, is the first order derivative of the heading error with respect to time; is the state variable change rate is the second order derivative of the lateral error with respect to time, is the second order derivative of the heading error with respect to time; u(t) is the control input, i.e., the front wheel steering angle δ f (t); is the disturbance caused by the known road curvature variation, is the desired heading rate; y(t) is the output state variable

[0022] wherein:

[0023]

[0024] Step 2: Obtain the extended state matrix considering the steering delay characteristics

[0025] Explicitly consider the dynamic characteristics of the steering system, and use a first-order lag system to parameterize the modeling of the steering delay characteristics to accurately reflect the dynamic behavior of the steering system. Let the target control input be the target front wheel steering angle δ pre , the actual actuator output steering angle be δ real , and their relationship be modeled as the following first-order lag system:

[0026]

[0027] wherein τ represents the time constant corresponding to the dynamic characteristics of the steering system, used to reflect the delay characteristics of the actuator.

[0028] Further discretize the relationship between the target front wheel steering angle and the actuator output steering angle under the sampling period T as follows:

[0029]

[0030] Introduce the first-order lag system into the single-track vehicle lateral dynamics model, further extend x(t) obtained in Step 1, add the current actual front wheel steering angle δ real and the target steering angle δ pre two state variables, and construct the following six-dimensional extended state vector:

[0031]

[0032] Combine the single-track vehicle lateral dynamics model and the extension module, and finally construct the discrete path tracking control model considering the steering delay characteristics as follows:

[0033]

[0034] wherein:

[0035]

[0036] Step 3: Obtain path tracking error model based on LPV module

[0037] Since tire cornering stiffness, vehicle mass and yaw moment of inertia are key structural parameters affecting the lateral dynamic model of monorail vehicles, and these parameters will dynamically change with vehicle load changes, tire wear, road condition changes, the following four time-varying parameters are selected to construct the LPV module: front wheel cornering stiffness C f , rear wheel cornering stiffness C r , vehicle mass m and vehicle yaw moment of inertia I z .

[0038] Based on the six-dimensional extended state vector, the LPV state space equation is established as follows:

[0039]

[0040] In the formula, ρ(k) is a time-varying parameter vector, ρ(k) = [C f C r m I z ] · ; A(μ(ρ(k)) is a system matrix related to time-varying parameters, reflecting the state evolution, control input and external disturbance influence of the vehicle under the current working condition, μ i (ρ(k)) is a weight coefficient dependent on time-varying parameters, satisfying Σμ i = 1, μ i ≥ 0; since there are inevitable measurement errors for each parameter, the weight coefficient is rewritten as

[0041] Since there are a large number of coefficients in the system matrix A(μ(ρ(k)) that are nonlinearly related to time-varying parameters, in order to facilitate online calculation and design of the controller, a convex polyhedron modeling method is used to approximate the LPV module. First, set the physical upper and lower bounds for each time-varying parameter:

[0042]

[0043] m∈[m min ,m max ]

[0044] I z ∈[I z min ,I z max ]

[0045] From the above four variable boundaries, 16 sets of vertex systems are generated represents all the extreme combinations of the time-varying space; for any state corresponding to ρ(t), the system matrix is represented as a weighted convex combination of the 16 vertex systems, as follows:

[0046]

[0047]

[0048] weighting coefficients are obtained by interpolating between the vertex values in real time, or are dynamically updated in conjunction with an online parameter estimator.

[0049] Step 4: Design the model predictive controller objective function

[0050] In order to more accurately depict the system dynamics and meet the modeling requirements of model predictive control, the current state ξ(k) and the control amount u k-1 are combined to form an extended state vector for subsequent control optimization calculations. The new state space model is represented as follows:

[0051]

[0052] wherein is an identity matrix of dimension N u , N u is the number of extended state vectors, N u = 6, and Δu(k) is the change in the control amount.

[0053] Iterative calculations are performed on the discrete path tracking control model to obtain the state equation and output state equation within the prediction horizon N p

[0054]

[0055] wherein:

[0056] X = [G(k+1) G(k+2) … G(k+N p )] T

[0057] Y = [η(k+1)η(k+2)…η(k+N p )] T

[0058]

[0059] wherein N p is the prediction horizon, and N c is the control horizon.

[0060] ​The quadratic function of the output state variable Y and the control input ΔU in the prediction time domain N p is defined as the objective function of the model predictive controller, and the extreme value constraint of the output state variable Y and the control input ΔU in the prediction time domain N p is defined as the constraint of the objective function:

[0061]

[0062] satisfies

[0063] wherein:

[0064] H = 2 (Θ T QΘ + R)

[0065] f T = 2 (Ψξ(k)) T QΘ + 2Γ T QΘ

[0066] wherein u min and u max are the maximum and minimum values of the system control input, i.e., the maximum and minimum values of the front wheel steering angle, Δu min and Δu max are the constraint maximum and minimum values of the input front wheel steering angle increment, η min and η max are the constraint maximum and minimum values of the output state variable.

[0067] The present application has the following beneficial effects:

[0068] 1. The present application is based on a single-rail vehicle lateral dynamics model, which can dynamically capture the changes of key parameters of the vehicle in the path tracking process, thereby providing a more accurate dynamic description of the system in the model predictive control process. Compared with traditional static linear models, the present application significantly improves the model prediction accuracy and the accuracy of the control response, ensuring the path tracking stability and control precision of the intelligent vehicle at high speed or in complex path changes.

[0069] 2. The present application addresses the problem of dynamic lag and time lag of the steering actuator, and introduces an extended state modeling and control increment constraint mechanism, which realizes effective modeling and feedforward compensation of the steering lag, not only improves the adaptability of the control system to the dynamic characteristics of the actual actuator, but also enhances the fault tolerance ability of the controller to the non-ideal execution lag behavior, ensuring the robustness and response coordination in the path tracking process. BRIEF DESCRIPTION OF DRAWINGS

[0070] Figure 1 is a schematic diagram of the present application.

[0071] Figure 2is a path tracking model schematic diagram of the present application considering uncertain disturbance and steering time delay.

[0072] Figure 3 is an implementation flowchart of the present application. DETAILED DESCRIPTION

[0073] The present application is further described below in conjunction with the accompanying drawings.

[0074] Figure 1 As shown, the composition of the intelligent vehicle time delay steering path tracking control system based on the LPV model of the present application specifically includes:

[0075] Extended state modeling module: in order to accurately depict the dynamic delay behavior of the steering actuator, the present application introduces a state variable related to the front wheel steering angle on the basis of the original four state variables, and a single-track vehicle lateral dynamics model with actuator response delay characteristics is constructed by modeling a first-order lag system.

[0076] Linear parameter time-varying module: the nonlinear path tracking error of the intelligent vehicle under different operating conditions is dynamically converted into a combination of multiple linear models. The selection of time-varying parameters is closely related to system stability and control accuracy, and specifically includes front wheel cornering stiffness, rear wheel cornering stiffness, vehicle mass, and vehicle yaw moment of inertia. By weighting and combining the constructed local linear models, and dynamically interpolating the real-time time-varying parameter trajectories, accurate modeling of the path tracking error dynamic system is achieved.

[0077] Model predictive controller: the single-track vehicle lateral dynamics model sends the real-time state information of the vehicle, including the real-time position information, heading information and speed information of the intelligent vehicle, to the tracking error dynamics model considering uncertain disturbance and steering time delay, and at the same time, the upper planning module also sends the position information, heading and curvature information of the expected path matching point to the tracking error dynamics model considering uncertain disturbance and steering time delay. Based on the tracking error dynamics model considering uncertain disturbance and steering time delay, the model predictive controller solves the quadratic programming problem to obtain the front wheel steering angle δ mpc .

[0078] As shown in Figures 1-3 , a kind of intelligent vehicle time delay steering path tracking control method based on LPV model, including the following steps:

[0079] Step 1: initialization

[0080] To realize intelligent vehicle steering control in complex dynamic environment, considering the lateral and yaw motion of the vehicle, a single-track vehicle lateral dynamics model is established.

[0081] According to Newton's law of motion, the force balance equation containing uncertain disturbance terms in the lateral and yaw directions is obtained:

[0082]

[0083] where F yf is the lateral force of the front wheels of the vehicle, F yr is the lateral force of the rear wheels of the vehicle.

[0084] Assuming that the tire slip angle is small, the front and rear tire lateral forces can be approximated as linear functions of the tire slip angle:

[0085]

[0086] The lateral dynamics model of the monorail vehicle is:

[0087]

[0088] Step 2: Obtain the extended state matrix considering the steering delay characteristics

[0089] According to the lateral dynamics model of the monorail vehicle and the position information, heading information and curvature at the matching point of the desired path, a path tracking error model considering the steering delay characteristics is designed as shown in Figure 2 The path tracking error model establishes the path tracking control relationship between the desired path and the monorail dynamics model of the intelligent vehicle. In the tracking error dynamics model considering multi-channel uncertainty disturbance, XOY is the global coordinate system, xoy is the vehicle coordinate system, τ r o r n r is the Frenet coordinate system.

[0090] Define the lateral error e d and the heading error as:

[0091]

[0092] where, and are the position vectors of the vehicle mass center o and the matching point o r of the desired path in the global coordinate system XOY, respectively; is the unit vector in the normal direction of the point o r ; ψ r is the desired heading angle; the lateral error e d is the lateral distance between the vehicle mass center o and the matching point o r ; the heading error is the difference between the actual yaw angle of the vehicle and the desired heading angle ψ r at the matching point o r .

[0093] Desired heading angle ψ r The first-order differential of can be approximately expressed as:

[0094]

[0095] Where k r For point o r The curvature of the road.

[0096] Lateral error e d and heading error And its derivative form is:

[0097]

[0098] Based on the single-track vehicle lateral dynamics model in step 1, the lateral error e is obtained d and heading error The expression for the second derivative of :

[0099]

[0100] Thus, the tracking error dynamic model is obtained:

[0101]

[0102] The delay characteristics are parameterized by using a first-order lag system to construct a

[0103] Path tracking control model After discretization, we get:

[0104]

[0105] By introducing the steering delay characteristic into the lateral dynamics model of the monorail vehicle, the original state variable set is further expanded to obtain the following extended state vector:

[0106]

[0107] Therefore, the discrete form of the path tracking model considering the steering delay characteristic is:

[0108]

[0109] Step 3: Obtain the path tracking error model based on the LPV module

[0110] Considering that tire cornering stiffness, vehicle mass and yaw inertia have significant influence on vehicle lateral dynamic response, and these parameters vary under different load, road condition and driving state, four time-varying parameters, front wheel cornering stiffness, rear wheel cornering stiffness, vehicle mass and vehicle yaw inertia, which have significant influence on system dynamics, are selected as the time-varying parameter vector ρ(k) of LPV:

[0111] ρ(k) = [C f C r m I z ] ·

[0112] The path tracking model in step 2 is converted to:

[0113] ξ(k+1) = A(μ(ρ(k)))ξ(k) + B(μ(ρ(k)))u(k)

[0114] The time-varying parameters can be obtained in real time through online parameter estimation or multi-source sensor measurement according to the current load state of the vehicle, the adhesion condition between the tire and the road and other operating condition information. The convex polyhedron modeling method is used to approximate the nonlinear relationship as a convex combination of multiple linear subsystems corresponding to the time-varying parameters:

[0115]

[0116] According to the number and discretization method of the selected time-varying parameters, 16 groups of vertex systems can be obtained by linearization in advance. By weighted convex combination of these vertex systems, a global LPV model covering the entire operating range can be constructed, so that the model predictive controller still has the ability of online adaptive adjustment in the presence of system parameter uncertainty and external disturbance, and realizes high-precision and strong-robustness path tracking control. Since there are inevitable errors in the actual measurement of time-varying parameters, the corresponding weight coefficients are modeled as uncertainties as follows:

[0117]

[0118] Combined with the above uncertainty modeling process, the global LPV state space model with time-varying parameter disturbance term is constructed as follows:

[0119]

[0120] Step 3: Design the model predictive controller objective function

[0121] The current state ξ(k) and the control amount u (k-1) are combined to construct a new system state vector Then the new state space model is

[0122] Prediction horizon is N p Control horizon is N c , and N c ≤ N p . The iteration calculation is performed on the discretized path tracking error model to obtain the state variable sequence within the prediction horizon N p

[0123]

[0124] Similarly, the output state variable sequence is:

[0125]

[0126] Based on the state variable sequence and the output state variable sequence, the state equation and the output equation within the prediction horizon N p

[0127]

[0128] To obtain the optimal control variable sequence, the quadratic function of the output state variable Y and the control input variable ΔU within the prediction horizon N p

[0129] J = Y(k) T QY(k) + ΔU T RΔU

[0130] In the formula, Q is the weight matrix of the output state variable Y, which is used to adjust the tracking effect of the model predictive controller on the desired path; R is the weight matrix of the control input variable ΔU, which is used to limit the amplitude of the control input variable.

[0131] Therefore, the model predictive optimal control problem can be expressed as a standard quadratic programming problem containing multiple constraint conditions:

[0132]

[0133] Step 4: According to steps 2 and 3, the front wheel steering angle control law can be designed as:

[0134] δ = δ + ΔU

[0135] Step 5: In the present application, for the final front wheel steering angle control variable obtained by solving in step 4, control constraint conditions are first applied to the control variable to meet the actual steering limit requirements of the front wheel of the intelligent vehicle. The front wheel steering angle instruction processed by the above-mentioned processing is used as the final control variable, which is transmitted to the actuator of the intelligent vehicle at a fixed update frequency.

[0136] ​​​The executor drives the front wheels to realize tracking control according to the received control signal, so that the intelligent automobile can run stably along the expected path under the actual working conditions with steering time delay and uncertain disturbance.

[0137] In addition, the control system continuously monitors the vehicle position information during the path tracking process to determine whether the intelligent automobile has reached the preset target position. If the determination result is "has reached", the control task is terminated, and the path tracking process is completed. If it has not been reached, steps 1 to 4 are re-executed to form a closed-loop iterative path tracking control process until the intelligent automobile successfully completes the scheduled path tracking task.

[0138] The present application is not limited to the details of the above-mentioned embodiments, and any equivalent concept or change within the technical scope disclosed in the present application is included in the protection scope of the present application.

Claims

1. A time-delay steering path tracking control method for an intelligent vehicle based on an LPV model, which utilizes an intelligent vehicle time-delay steering path tracking control system for control, characterized in that: The intelligent vehicle time-delay steering path tracking control system includes an initialization module, an extended system model construction module, a linear parameter time-varying module, and a model predictive controller; the intelligent vehicle time-delay steering path tracking control system is hereinafter referred to as the control system, the extended system model construction module is hereinafter referred to as the extension module, and the linear parameter time-varying module is hereinafter referred to as the LPV module; The control method comprises the following steps: Step 1: Initialization; Step 2: Obtain the extended state matrix considering the steering delay characteristics; Step 3: Obtain the path tracking error model based on the LPV module; Step 4: Design the model predictive controller objective function.

2. The method for controlling time-delay steering path tracking of an intelligent vehicle based on an LPV model according to claim 1, characterized in that: The initialization module is responsible for checking whether the signal transmission and reception of the perception module, mapping and positioning module, decision and planning module, and actuator are normal, loading the parameters of the monorail vehicle lateral dynamics model, establishing the vehicle coordinate system, loading the tracking error dynamics model parameters, and loading the model predictive controller parameters; the perception module, mapping and positioning module, and decision and planning module are universal modules outside the control system, and work together with the control system to ensure the normal operation of the smart car; The extended system model building module builds an extended state space model based on the tracking error dynamics model; introduces a steering angle state variable based on the original system state variable, and uses a first-order lag system to model the actual front wheel angle, thereby forming an extended system model that includes steering dynamics; During vehicle operation, the linear parameter time-varying module dynamically generates a state space model required for predictive control based on the vehicle's current state variables. The linear parameter time-varying module uses the front wheel cornering stiffness, rear wheel cornering stiffness, vehicle mass, and vehicle yaw moment of inertia as time-varying parameters, combined with the current vehicle longitudinal speed, and adopts a multi-model weighted approach to online construct a time-varying system state matrix and input matrix. The model predictive controller solves the quadratic programming problem based on the expansion module and the LPV module to obtain the optimal front wheel angle control law.

3. The method for controlling time-delay steering path tracking of an intelligent vehicle based on an LPV model according to claim 1, characterized in that: The steps of initialization described in step 1 are as follows: The initialization module is responsible for checking whether the signal communication status of the perception module, mapping and positioning module, decision and planning module, and actuator is normal; loading the parameters of the monorail vehicle lateral dynamics model, model predictive controller parameters, and feedforward compensator parameters; and receiving the position information, heading angle information, and road curvature information of the path matching point sent by the decision and planning module. The path matching point is the projection point of the current intelligent vehicle position on the desired path, that is, the point closest to the desired path. The vehicle coordinate system is established as follows: with the center of mass of the vehicle as the origin, the positive direction of the x-axis points to the vehicle's forward direction, the positive direction of the y-axis points to the left side of the vehicle, and the z-axis is vertically upward, forming a right-handed rectangular coordinate system together with the x-axis and y-axis; The lateral dynamics model of the monorail vehicle is as follows: Where C f and C r are the cornering stiffness of the front and rear wheels of the vehicle respectively; m is the mass of the vehicle; v x is the longitudinal velocity of the vehicle; y is the lateral displacement of the vehicle coordinate system, is the velocity in the lateral direction of the vehicle coordinate system, is the acceleration in the lateral direction of the vehicle coordinate system; is the vehicle yaw angle, is the vehicle yaw angular velocity, is the vehicle yaw angular acceleration; δ f is the front wheel turning angle of the vehicle; I z is the moment of inertia of the vehicle's center of mass about the Z axis, which forms a right-handed system with the X and Y axes in the vehicle coordinate system; a and b are the distances from the vehicle's center of mass to the front and rear axles, respectively; Based on the position information, heading information and curvature of the desired path matching point, the tracking error dynamics model considering multi-channel uncertainty disturbance is established as follows: Where x(t) is the state variable in vehicle path tracking control, and x(t) is e d is the lateral error, is the first-order derivative of the lateral error with respect to time, is the heading error, is the first-order derivative of the heading error with respect to time; is the rate of change of state variables is the second-order derivative of the lateral error with respect to time, is the second-order derivative of the heading error with respect to time; u(t) is the control input, which is the front wheel steering angle δ of the vehicle f (t); is the disturbance caused by the known change in road curvature, is the desired heading angular velocity; y(t) is the output state variable in:

4. The method for controlling time-delay steering path tracking of an intelligent vehicle based on an LPV model according to claim 1, characterized in that: The method for obtaining the extended state matrix considering the steering delay characteristic in step 2 is as follows: The dynamic characteristics of the steering system are explicitly considered, and the steering delay characteristics are parameterized and modeled using a first-order lag system to accurately reflect the dynamic behavior of the steering system. Assume that the target control input is the target front wheel steering angle δ pre , the actual actuator output angle is δ real , the relationship between the two is modeled as the following first-order lag system: Wherein, τ represents the time constant corresponding to the dynamic characteristics of the steering system, which is used to reflect the delay characteristics of the actuator; The relationship between the target front wheel angle and the actuator output angle is further discretized at the sampling period T as follows: The first-order hysteresis system is introduced into the lateral dynamics model of the single-track vehicle, and the x(t) obtained in step 1 is further expanded to increase the current actual front wheel angle δ real and the target angle δ pre Two state variables, construct the following six-dimensional extended state vector: Combining the monorail vehicle lateral dynamics model with the extended module, a discrete path tracking control model that takes into account the steering delay characteristics is finally constructed as follows: in:

5. The time-delay steering path tracking control method for an intelligent vehicle based on an LPV model according to claim 1, characterized in that: The method for obtaining the path tracking error model based on the LPV module in step 3 is as follows: Since tire cornering stiffness, vehicle mass and yaw moment of inertia are key structural parameters affecting the lateral dynamics model of monorail vehicles, and these parameters will change dynamically with changes in vehicle load, tire wear and road conditions, the following four time-varying parameters are selected to construct the LPV module: front wheel cornering stiffness C f , rear wheel cornering stiffness C r , vehicle mass m and vehicle yaw moment of inertia I z ; The state space equation of the LPV module is established based on the six-dimensional extended state vector as follows: Where ρ(k) is a time-varying parameter vector, ρ(k)=[C f C r m I z ] · ; A(μ(ρ(k)) is the system matrix related to the time-varying parameters, reflecting the state evolution, control input and external disturbance influence of the vehicle under the current working conditions, μ i (ρ(k)) is the weight coefficient that depends on the time-varying parameter and satisfies ∑μ i =1,μ i ≥0; Since each parameter inevitably has measurement errors, the weight coefficient is rewritten as Since the system matrix A(μ(ρ(k)) contains a large number of coefficients that are nonlinearly related to time-varying parameters, in order to facilitate online calculation and design of the controller, the LPV module is approximated by a convex polyhedron modeling method. First, the physical upper and lower bounds are set for each time-varying parameter: m∈[m min ,m max ] I z ∈[I z min ,I z max ] The above four variable boundaries are combined to generate 16 groups of vertex systems Represents all extreme combinations of the time-varying space; for the state corresponding to any moment ρ(t), the system matrix is ​​represented as a weighted convex combination of 16 vertex systems, as follows: Weight coefficient It is obtained by interpolating real-time time-varying parameter values ​​between vertices, or dynamically updated in combination with an online parameter estimator.

6. The time-delay steering path tracking control method for an intelligent vehicle based on an LPV model according to claim 1, characterized in that: The method for designing the objective function of the model predictive controller described in step 4 is as follows: In order to more accurately describe the system dynamics and meet the modeling requirements of model predictive control, the current state ξ(k) is combined with the control quantity u at the previous moment. k-1 Joint construction to form an extended state vector For subsequent control optimization calculations; the new state space model is expressed as follows: in The dimension is N u The identity matrix, N u is the number of extended state vectors, N u =6, Δu(k) is the change of the control quantity; The discrete path tracking control model is iteratively calculated to obtain the predicted time domain N p The internal state equation and output state equation are as follows: Where: X=[G(k+1)G(k+2)…G(k+N p )] T Y=[η(k+1)η(k+2)…η(k+N p )] T Among them, N p is the prediction time domain, N c To control the time domain; The prediction time domain N p The quadratic function of the output state variable Y and the control input ΔU is defined as the objective function of the model predictive controller, and the prediction time domain N p The maximum value constraints of the output state variable Y and the control input ΔU are defined as the constraints of the objective function: satisfy Where: H=2(Θ T QΘ+R) f T =2(Ψξ(k)) T QT+2G T QT Where u min and u max are the maximum and minimum values ​​of the system control input, that is, the maximum value of the front wheel angle, Δu min and Δu max are the maximum and minimum constraints of the input front wheel angle increment, η min and η max are the constrained maximum values ​​of the output state variables respectively.

Images