A commercial vehicle transverse layered robust predictive control method and related device

Abstract

The application discloses a kind of commercial vehicle transverse layered robust predictive control method and related device, it is related to vehicle control technical field, which comprises obtaining vehicle basic information and constructing continuous-time dynamics model, linearization, discretization processing and nominal state recursion are carried out, and discrete-time nominal recursion equation is determined;According to discrete nominal lateral state vector and vehicle real discrete lateral state vector, error state update equation is determined, and lower layer robust feedback gain is solved, feedback compensation and gain matrix are obtained, error robust positive invariant set is constructed in combination with error state update equation, the control input constraint set and state constraint set after tightening are determined, then in combination with discrete-time nominal recursion equation, nominal control quantity is determined;According to feedback compensation and nominal control quantity, total control quantity is obtained.The application can guarantee the stability and tracking accuracy of control, eliminate the target conflict of upper and lower controllers, realize safe and smooth lateral trajectory tracking.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and in particular to a lateral hierarchical robust predictive control method and related devices for commercial vehicles. Background Technology

[0002] Compared to passenger vehicles, commercial vehicles exhibit significantly more complex and time-varying dynamic characteristics in actual operation. Commercial vehicles face lateral control challenges at high speeds, as their dynamic characteristics become highly time-varying due to wide variations in load. Furthermore, their large lateral area makes them susceptible to severe physical disturbances from strong crosswinds and uneven road surfaces, highlighting the significant shortcomings of traditional control schemes. Traditional single-layer model predictive control (MPC) is highly dependent on model accuracy, and under severe model mismatch or strong disturbances, it can easily lead to unsolvable optimization problems and control failure. Traditional single-layer robust control (such as...) While conventional control methods can suppress disturbances, they are fundamentally unable to handle key physical constraints such as steering angle and speed, and are also difficult to integrate with high-precision maps for predictive tracking of forward-looking path curvature. Existing Tube Model Predictive Control (Tube MPC) or Min-Max Model Predictive Control (Min-Max MPC) methods, although theoretically combining robustness and predictive control, require excessively large disturbance boundaries to meet theoretical safety requirements when facing the "strong disturbances" of commercial vehicles. This leads to overly tight constraints, a drastically reduced feasible region, heavy computational burden, and a high risk of unsolvable problems. In the conventional "upper-layer MPC + lower-layer robust compensation" architecture, the inconsistency between the upper and lower layer objective functions can easily cause commands to cancel each other out or superimpose, resulting in high-frequency steering wheel vibration at high speeds, severely impacting driving safety and ride comfort.

[0003] Therefore, there is an urgent need to propose a lateral hierarchical robust predictive control method for commercial vehicles that can simultaneously ensure control stability and tracking accuracy from the mathematical and system architecture levels under high-speed and harsh operating conditions with large modeling errors, complex unmodeled dynamics and strong external disturbances, and effectively eliminate target conflicts between upper and lower level controllers to achieve safe and smooth lateral trajectory tracking. Summary of the Invention

[0004] The purpose of this application is to provide a lateral layered robust predictive control method and related device for commercial vehicles, which can ensure the stability and tracking accuracy of control, eliminate target conflicts between upper and lower level controllers, and achieve safe and smooth lateral trajectory tracking.

[0005] To achieve the above objectives, this application provides the following solution: Firstly, this application provides a lateral hierarchical robust predictive control method for commercial vehicles, including: Obtain basic vehicle information; the basic vehicle information includes sampling period, current longitudinal speed of the vehicle, reference path information, vehicle's true discrete lateral state vector and nominal vehicle parameters.

[0006] Based on the vehicle's basic information, a continuous-time dynamics model is constructed, and linearization, discretization, and nominal state recursion are performed on the continuous-time dynamics model to determine the discrete-time nominal recursive equation.

[0007] Based on the discrete nominal lateral state vector in the discrete-time nominal recursive equation and the vehicle's real discrete lateral state vector, the error state update equation is determined.

[0008] Based on the aforementioned error state update equation, the lower layer is discretely solved. Robust feedback gain is used to obtain the lower-level robust feedback gain matrix and the feedback compensation amount generated by the lower-level robust feedback control.

[0009] Based on the lower-level robust feedback gain matrix and the error state update equation, an error robust positive invariant set is constructed, and the tightened control input constraint set and the tightened state constraint set are determined according to the error robust positive invariant set.

[0010] Based on the discrete-time nominal recursive equation, the tightened set of control input constraints, and the tightened set of state constraints, the nominal control quantity generated by the upper-level nominal model predictive control programming is determined.

[0011] Based on the feedback compensation amount and the nominal control amount, the total control amount of the actual output is obtained; the total control amount is used as the input amount of the vehicle steering actuator, so that the steering actuator performs steering action according to the total control amount.

[0012] Secondly, this application provides a lateral hierarchical robust predictive control device for commercial vehicles, including a vehicle control mechanism and a steering actuator.

[0013] The vehicle control mechanism is used to perform the steps of implementing the above-described lateral hierarchical robust predictive control method for commercial vehicles, and outputs the total control quantity.

[0014] The steering actuator is connected to the vehicle control mechanism and is used to receive the total control quantity and perform steering actions according to the total control quantity.

[0015] Thirdly, this application provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described commercial vehicle lateral hierarchical robust predictive control method.

[0016] According to the specific embodiments provided in this application, this application has the following technical effects: This application provides a lateral hierarchical robust predictive control method and related devices for commercial vehicles. By acquiring basic vehicle information and determining discrete-time nominal recursive equations, it solves the model mismatch problem caused by load changes and crosswind interference in commercial vehicles, thereby improving the lateral stability of the vehicle under harsh conditions. By constructing error state update equations and solving the lower-level robust feedback gain offline, it avoids complex online calculations and solves the problems of insufficient computing power and poor real-time performance of onboard chips, achieving millisecond-level fast response of the controller and meeting automotive-grade deployment requirements. By constructing an error robust positive invariant set and calculating a tightened constraint set, and deducting safety margins to divide virtual constraints... By tightening the constraint set to solve the shortcomings of traditional robust control constraints being conservative and having no optimization solution, the system achieves recursive feasibility of constraints during vehicle operation, eliminating lane departure and steering saturation problems. By combining the tightened constraint set to solve the upper-level nominal control quantity and matching the upper and lower-level control logic, the system solves the pain points of inconsistent objectives and internal friction caused by mutual game between instructions in traditional hierarchical control, achieving alignment of upper and lower-level control objectives. Finally, the system superimposes the feedback compensation quantity and the nominal control quantity to generate the total control quantity, optimizes the control instructions of the steering actuator, effectively eliminates the target conflict between upper and lower level controllers, significantly improves the accuracy of vehicle trajectory tracking and the stability of control, and achieves safe and smooth lateral trajectory tracking. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is an application environment diagram of a lateral hierarchical robust predictive control method for commercial vehicles according to one embodiment of this application; Figure 2 A flowchart illustrating a lateral hierarchical robust predictive control method for commercial vehicles provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of a computer device provided in an embodiment of this application. Detailed Implementation

[0019] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0020] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, the application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0021] The lateral hierarchical robust predictive control method for commercial vehicles provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be set up independently, integrated into server 104, or placed in the cloud or on another server. Terminal 102 can send basic vehicle information to server 104. After receiving the basic vehicle information, server 104 constructs a continuous-time dynamics model based on the basic vehicle information, and performs linearization, discretization, and nominal state recursion on the continuous-time dynamics model to determine the discrete-time nominal recursive equation; based on the discrete nominal lateral state vector in the discrete-time nominal recursive equation and the vehicle's actual discrete lateral state vector, it determines the error state update equation; based on the error state update equation, it discretely solves the lower-level equation. Robust feedback gain is used to obtain the lower-level robust feedback gain matrix and the feedback compensation amount generated by the lower-level robust feedback control. Based on the lower-level robust feedback gain matrix and the error state update equation, an error robust positive invariant set (RPI) is constructed, and the tightened control input constraint set and the tightened state constraint set are determined according to the error robust positive invariant set. Based on the discrete-time nominal recursive equation, the tightened control input constraint set and the tightened state constraint set, the nominal control amount generated by the upper-level nominal model predictive control planning is determined. According to the feedback compensation amount and the nominal control amount, the actual output total control amount is obtained. The server 104 can feed back the obtained actual output total control amount to the terminal 102. In addition, in some embodiments, the commercial vehicle lateral hierarchical robust predictive control method can also be implemented by the server 104 or the terminal 102 alone. For example, the terminal 102 can directly calculate the actual output total control amount based on the vehicle basic information, or the server 104 can obtain the vehicle basic information from the data storage system and calculate the actual output total control amount based on the vehicle basic information.

[0022] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, and smart in-vehicle devices. Portable wearable devices can include smartwatches, smart bracelets, and head-mounted devices. The server 104 can be implemented using a standalone server or a server cluster composed of multiple servers, or it can be a cloud server.

[0023] In one exemplary embodiment, such as Figure 2 As shown, a lateral hierarchical robust predictive control method for commercial vehicles is provided. This method is executed by computer equipment, specifically by a terminal or server alone, or by both a terminal and a server. In this embodiment, the method is applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 207.

[0024] Step 201: Obtain basic vehicle information; the basic vehicle information includes sampling period, current longitudinal speed of the vehicle, reference path information, vehicle's true discrete lateral state vector, and nominal vehicle parameters.

[0025] Step 202: Based on the vehicle basic information, construct a continuous-time dynamics model, and perform linearization, discretization and nominal state recursion on the continuous-time dynamics model to determine the discrete-time nominal recursive equation.

[0026] Step 203: Determine the error state update equation based on the discrete nominal lateral state vector in the discrete-time nominal recursive equation and the vehicle's real discrete lateral state vector.

[0027] Step 204: Based on the error state update equation, discretely solve the lower layer... Robust feedback gain is used to obtain the lower-level robust feedback gain matrix and the feedback compensation amount generated by the lower-level robust feedback control.

[0028] Step 205: Based on the lower-level robust feedback gain matrix and the error state update equation, construct the error robust positive invariant set, and determine the tightened control input constraint set and the tightened state constraint set according to the error robust positive invariant set.

[0029] Step 206: Based on the discrete-time nominal recursive equation, the tightened set of control input constraints, and the tightened set of state constraints, determine the nominal control quantity generated by the upper-level nominal model predictive control programming.

[0030] Step 207: Based on the feedback compensation amount and the nominal control amount, the total control amount of the actual output is obtained; the total control amount is used as the input amount of the vehicle steering actuator, so that the steering actuator performs steering action according to the total control amount.

[0031] It should be noted that the lateral hierarchical robust predictive control method for commercial vehicles proposed in this application can also be executed by a vehicle controller (vehicle control mechanism), domain controller, autonomous driving control unit, or other electronic devices with data processing and actuator control functions. In the embodiments of this application, it is preferably executed by the vehicle controller.

[0032] By implementing steps 201 to 207 above, this application can guarantee control stability and tracking accuracy, eliminate target conflicts between upper and lower level controllers, and achieve safe and smooth lateral trajectory tracking. Furthermore, this application can also employ an upper-level nominal MPC responsible for globally optimal forward tracking and constraint processing, while the lower level is based on an error system... The feedback layered architecture is responsible for local robust stability. This layered robust architecture, adapted to harsh operating conditions, is specifically designed for high-speed, high-interference scenarios. Simultaneously, based on the robust positive invariant set of the error system under the worst-case disturbance, the tightened set of control input constraints and the tightened set of state constraints are obtained offline analytically. This mechanism, based on the recursive feasibility of the robust positive invariant set, ensures that as long as the upper-level nominal MPC has a solution within the tightened boundary, the real vehicle will never exceed the boundary risk when subjected to strong disturbances, thus guaranteeing the recursive feasibility of the system without increasing the online computational load.

[0033] It should be noted that the vehicle basic information obtained in step 201 is the input quantity acquired or determined by the vehicle controller (i.e., acquiring the input quantity and determining the sampling parameters). The vehicle basic information includes the sampling period. Current longitudinal speed of the vehicle Reference path information, vehicle's true discrete lateral state vector and nominal vehicle parameters Wherein, the vehicle's current longitudinal speed The reference path information is obtained from vehicle sensors or output by a vehicle state estimation algorithm. The reference path information includes at least the information obtained by the vehicle at the sampling time. and the reference path curvature sequence in the subsequent prediction time domain It may further include the corresponding reference lateral state sequence. ;in, For discrete sampling sequence number, For the first Each sampling time; For vehicles at the sampling time The discrete reference path curvature; For vehicles at the sampling time The discrete reference path curvature; For the vehicle in the Each sampling time The reference lateral state vector; For the vehicle in the Each sampling time The reference lateral state vector; the reference path information is output by the path planning module or calculated by the high-precision map and positioning module. For the vehicle in the Each sampling time The measured and / or estimated values ​​of the vehicle's true discrete lateral state vector, wherein the lateral state vector includes at least the vehicle's lateral position error. Vehicle heading angle error lateral velocity of vehicle center of gravity and vehicle yaw rate One or more of the following. The nominal vehicle parameters. At least including vehicle mass Moment of inertia of the vehicle about its vertical axis Distance from vehicle center of gravity to front axle Distance from vehicle center of gravity to rear axle Equivalent lateral stiffness of the vehicle's front axle Equivalent lateral stiffness of the rear axle of the vehicle The number of nominal vehicle parameters will vary depending on the vehicle model and dynamics model. The specific values ​​can be roughly defined based on engineering experience.

[0034] In another exemplary embodiment of this application, step 202 is replaced by steps 301 to 304: Step 301: Based on the two-dimensional monorail model and the vehicle's basic information, construct a continuous-time dynamic model.

[0035] This application embodiment requires establishing a continuous-time dynamics model (continuous-time commercial vehicle lateral dynamics model) and defining a lateral state vector. The vehicle controller establishes a continuous-time dynamics model based on a two-dimensional monorail model and the aforementioned vehicle basic information, and defines the expression for the lateral state vector as follows: (1); in, This is the horizontal state vector; It is the transpose of the vector.

[0036] The vehicle controller defines the expression for the control input as follows: (2); in, u For control input; This is the input for the front wheel steering angle or equivalent steering control.

[0037] The vehicle controller models the reference path curvature as an expression for a measurable known disturbance as follows: (3); in, For measurable known disturbances; The reference path curvature.

[0038] The expression for the constraint conditions satisfied by the lateral force balance and yaw moment balance of the vehicle is: (4); in, For vehicle quality; This represents the vehicle's current longitudinal speed. The lateral velocity of the vehicle's center of gravity; Let be the first derivative of the lateral velocity of the vehicle's center of mass with respect to time; The vehicle's yaw rate; Let be the first derivative of the vehicle's yaw rate with respect to time; Let be the moment of inertia of the vehicle about its vertical axis; This is the equivalent lateral force on the front axle; This is the equivalent lateral force on the rear axle; This is the distance from the vehicle's center of gravity to the front axle. This is the distance from the vehicle's center of gravity to the rear axle. This is the equivalent lateral disturbance force; This is the equivalent yaw disturbance moment.

[0039] Under the linear tire assumption, the constraint conditions satisfied by the front and rear axle lateral forces are expressed as follows: (5); in, This is the equivalent lateral stiffness of the vehicle's front axle. This is the equivalent lateral stiffness of the rear axle of the vehicle.

[0040] The expressions for the constraints that the lateral position error and the heading angle error must satisfy are as follows: (6); in, This is the first derivative of the vehicle's lateral position error with respect to time. This is the first derivative of the vehicle heading angle error with respect to time.

[0041] Step 302: Linearize the continuous-time dynamics model to obtain the continuous-time nominal state-space equation.

[0042] It should be noted that the vehicle controller is based on nominal vehicle parameters. Linearizing the continuous-time dynamics model yields the continuous-time nominal state-space equations. The expression for these equations is: (7); in, For the continuous-time nominal state-space equation in continuous time The possible values ​​of ; For the horizontal state vector in continuous time The value of ; For continuous-time control input; For continuous-time reference path curvature; For continuous external physical disturbances; The nominal state matrix is ​​a continuous-time matrix; The nominal input matrix is ​​for continuous time; The nominal reference input matrix is ​​a continuous-time matrix; Assign matrices to the continuous-time perturbations. The expressions for each of the above matrices are shown in equations (8), (9), (10), and (12), respectively.

[0043] It should be noted that the nominal vehicle parameters mentioned above... Refers to vehicle mass Moment of inertia of the vehicle about its vertical axis Distance from vehicle center of gravity to front axle Distance from vehicle center of gravity to rear axle Equivalent lateral stiffness of the vehicle's front axle Equivalent lateral stiffness of the rear axle of the vehicle The parameter set consists of the nominal values. In the embodiments of this application, the equations or matrices marked with "nominal" are based on these nominal vehicle parameters. The corresponding expression obtained. The vehicle controller combines equations (4) to (6) to obtain the expression based on the nominal vehicle parameters. The expression for the continuous-time nominal state-space matrix is: (8).

[0044] The expression for the continuous-time nominal input matrix is: (9).

[0045] The expression for the continuous-time nominal reference input matrix is: (10).

[0046] In one embodiment, the vehicle controller defines the expression for continuous-time external physical disturbances as follows: (11); in, For continuous time equivalent lateral disturbance force; It is the continuous-time equivalent yaw disturbance moment.

[0047] Accordingly, the expression for the continuous-time perturbation allocation matrix is: (12).

[0048] Step 303: The continuous-time nominal state-space equation is discretized using the explicit Euler discretization method to obtain the discrete-time dynamic model.

[0049] It should be noted that the vehicle controller is set to the first... Each sampling time is ,in, The sampling period; For discrete sampling sequence numbers; For the first Each sampling time.

[0050] The vehicle controller defines discrete variables that satisfy , , , ,in, For the continuous-time lateral state vector at the th The values ​​taken at each sampling time; For continuous time control input at the first The values ​​taken at each sampling time; For continuous-time reference path curvature input at the 1st The values ​​taken at each sampling time; For the continuous-time external physical disturbance vector at the th... The values ​​taken at each sampling time; For vehicles at the sampling time The discrete lateral state vector; For vehicles at the sampling time Discrete control input; For vehicles at the sampling time The discrete external physical disturbance vector.

[0051] In one embodiment of this application, the vehicle controller adopts the zero-order hold assumption, within the interval Internal retention and Constant, where, Indicates from the first From the sampling time to the... The sampling interval prior to each sampling time point. The vehicle controller uses the explicit Euler discretization method to discretize equation (7) to obtain the discrete-time dynamic model. The expression of the discrete-time dynamic model is: (13); in, For vehicles at the sampling time The discrete lateral state vector; The continuous-time horizontal state vector at the sampling time The first derivative with respect to time.

[0052] Step 304: Based on the discrete-time dynamics model, determine the discrete-time nominal dynamics model, and perform nominal state recursion based on the discrete-time nominal dynamics model to determine the discrete-time nominal recursion equation.

[0053] It should be noted that substituting equation (7) into equation (13) yields the result based on nominal vehicle parameters. The constructed discrete-time nominal dynamics model is the discrete-time state-space equation. The expression for the discrete-time nominal dynamics model is: (14); in, Let be the nominal discrete state matrix. For the nominal discrete input matrix, The nominal discrete reference input matrix, Assign a matrix to the discrete-time perturbation.

[0054] (15); in, The identity matrix that matches the corresponding matrix dimension.

[0055] It needs to be clarified that in equation (13) This is the general notation for the discrete lateral state vector; in the subsequent nominal prediction process, it is used... This represents the discrete nominal lateral state vector; when describing the actual operating state of the vehicle, it uses... This represents the vehicle's true discrete lateral state vector.

[0056] In one exemplary embodiment of this application, by defining a hierarchical control architecture, performing nominal state recursion, and determining an error state update model, a foundation is laid for subsequent offline solving of the lower layer. Robust Feedback Gain The offline construction of the error robust positive invariant set and determination of constraint tightening, as well as the construction of the upper-level nominal model predictive control optimization problem, provide a physical foundation. To achieve coordination between the upper-level model predictive control and the lower-level robust feedback control, the vehicle controller adopts a hierarchical control architecture. The vehicle controller defines the vehicle at the sampling time... The expression for the total control quantity actually output to the steering actuator (steering mechanism) is: (16); in, For vehicles at the sampling time The total control quantity actually output to the steering actuator; For vehicles at the sampling time Nominal control quantities obtained from the control program predicted by the upper-level nominal model; For vehicles at the sampling time The feedback compensation amount generated by the lower-level robust feedback control.

[0057] To perform deterministic state extrapolation in predictive control, the vehicle's MPC controller ignores unknown future disturbances and employs a discrete-time nominal dynamics model for nominal state recursion, thus determining the discrete-time nominal recursive equation. The expression for the discrete-time nominal recursive equation is as follows: (17); in, For vehicles at the sampling time The nominal lateral state vector; For vehicles at the sampling time The nominal lateral state vector; The nominal discrete state matrix; The nominal discrete input matrix; Let be the nominal discrete reference input matrix.

[0058] In another exemplary embodiment of this application, step 203 is replaced by steps 401 to 404: Step 401: Calculate the difference between the discrete nominal lateral state vector in the discrete-time nominal recursive equation and the vehicle's actual discrete lateral state vector to obtain the error state vector.

[0059] It should be noted that, in actual operation, the vehicle is affected by parameters such as vehicle load changes, tire wear, and external physical disturbances such as crosswinds. The update process of the vehicle's true discrete lateral state vector can be described by the following expression.

[0060] (18); in, For vehicles at the sampling time The true lateral state vector; For vehicles at the sampling time The true lateral state vector; The true state matrix includes perturbations with unknown parameters; The true input matrix includes perturbations with unknown parameters; It is the true reference input matrix that contains perturbations with unknown parameters.

[0061] Based on the vehicle's real discrete lateral state vector and discrete nominal lateral state vector, the error state vector defined by the vehicle controller is obtained; the expression for the error state vector is: (19); in, For vehicles at the sampling time The error state vector is used to characterize the difference between the vehicle's true lateral state vector and the discrete nominal lateral state vector.

[0062] Step 402: Based on the error state vector, determine the error state vector at the next moment.

[0063] It should be noted that, in order to obtain the evolution law of the error state, subtracting equation (17) from equation (18) and performing algebraic identity transformation in conjunction with equation (16) yields the error state vector at the next moment. The expression for the error state vector at the next moment is: (20); in, For vehicles at the sampling time The error state vector.

[0064] Step 403: Define the parameter mismatch term, unmodeled dynamic influence term, and external physical disturbance term in the error state vector at the next moment as discrete generalized disturbances.

[0065] It should be noted that, in order to facilitate the design of the lower-level robust controller, the vehicle controller extracts the parameter mismatch term, the unmodeled dynamic influence term, and the external physical disturbance term in equation (20) and equates them to discrete generalized disturbances. The expression for the discrete generalized perturbation is: (twenty one); in, For vehicles at the sampling time The discrete generalized perturbation is used to uniformly characterize the parameter mismatch term, unmodeled dynamic influence term, and external physical perturbation term between the discrete-time nominal recursive equation and the real vehicle.

[0066] Step 404: Determine the error state update equation based on the generalized perturbation allocation matrix and the discrete generalized perturbation.

[0067] Furthermore, a generalized perturbation allocation matrix is ​​introduced. Then equation (20) can be simplified as follows to determine the error state update equation. The expression of the error state update equation is: (twenty two); in, Assign a matrix to the generalized perturbation to characterize the discrete generalized perturbation. Acting on the error state vector The input channel.

[0068] It should be noted that the discrete-time perturbation allocation matrix only represents the input channel of discrete external physical perturbations, while the generalized perturbation allocation matrix here... This characterizes the input channel of a discrete generalized perturbation, which integrates parameter mismatch terms, unmodeled dynamic influence terms, and external physical perturbation terms as discrete generalized perturbations. In a preferred embodiment, since parameter mismatch terms may directly affect all dimensions of the state vector, a generalized perturbation allocation matrix can also be set. For the error state vector Dimension matching identity matrix .

[0069] In another exemplary embodiment of this application, step 204 is replaced by steps 501 to 505: Step 501: The lower-level robust feedback controller is subjected to state feedback, and the error closed-loop system is obtained based on the error state update equation.

[0070] It should be noted that the vehicle controller uses state feedback for the lower-level robust feedback controller, and the expression for the state feedback is as follows: (twenty three); in, K This is the feedback gain matrix.

[0071] Substituting equation (23) into the error state update equation (21), we obtain the error closed-loop system. The expression for the error closed-loop system is: (twenty four); Step 502: Preset the state weight matrix and control weight matrix, and construct the lower-level performance output.

[0072] In order to achieve predictive control of the upper nominal model and the lower level Cost alignment between robust feedback controls, with the vehicle controller pre-setting a shared state weight matrix. and control weight matrix and the state weight matrix and control weight matrix It is used simultaneously for constructing the lower-level performance output and calculating the upper-level nominal model predictive control cost function. Based on this, the vehicle controller defines the expression for the performance output as follows: (25); in, For vehicles at the sampling time Performance output; For shared state weight matrix; For shared control weight matrix.

[0073] Equivalently, equation (25) can be written as: (26); in, The error state weighting matrix; This is the weighted matrix for feedback compensation.

[0074] The expression for the error state weighting matrix is: (27).

[0075] The expression for the weighted matrix of the feedback compensation amount is: (28).

[0076] Step 503, Setting Performance indicators, determined The constraints.

[0077] It should be noted that the vehicle controller settings The performance requirements stipulate that, under zero initial conditions, the closed-loop error system should recover from the discrete generalized disturbance. To performance output transfer function of The following constraints must be met.

[0078] The The expression for the constraint condition is: (29); in, Let be the transfer function of the closed-loop error system from the discrete generalized perturbation to the lower-level performance output. This is a preset upper bound for robust performance.

[0079] Step 504: Using the discrete-time bounded real lemma and Lyapunov stability theory, the constraints are transformed into linear matrix inequalities.

[0080] Step 505: Solve the linear matrix inequality offline using the interior point method to determine the lower-level robust feedback gain matrix and the feedback compensation amount generated by the lower-level robust feedback control.

[0081] In another exemplary embodiment of this application, step 504 is replaced by steps 601-602: Step 601: Set the Lyapunov matrix and define the symmetric positive definite matrix, and introduce auxiliary variables to obtain the constraint conditions of the feedback gain matrix.

[0082] It should be noted that, in order to... The performance constraints are transformed into a solvable convex optimization problem, and the vehicle controller is transformed based on the discrete-time bounded real lemma and Lyapunov stability theory. Let the Lyapunov matrix be... To eliminate errors in the closed-loop system and The nonlinear coupling terms resulting from multiplication are defined by an auxiliary matrix. And introduce auxiliary variables The constraint conditions for the feedback gain matrix are obtained. The expression for the constraint conditions of the feedback gain matrix is: (30); in, Y The auxiliary matrix is ​​a symmetric positive definite matrix. L As an auxiliary variable; Y -1 It is the inverse of the auxiliary matrix.

[0083] Step 602, combining the Schul complement lemma and congruent transformation, the... The constraints are transformed into linear matrix inequalities.

[0084] It should be noted that the expression for the linear matrix inequality is: (31).

[0085] The vehicle controller solves equation (31) offline using the interior-point method to obtain a solution that satisfies the requirements of robust stability and disturbance suppression. , and the corresponding preset robust performance upper bound And the lower-level robust feedback gain matrix is ​​recovered according to equation (30). This is for use in subsequent constraint tightening calculations and online feedback compensation.

[0086] In another exemplary embodiment of this application, step 205 is replaced by steps 701 to 704: Step 701: Based on the lower-level robust feedback gain matrix and the error state update equation, construct the error robust positive invariant set.

[0087] Based on the lower-level robust feedback gain matrix and auxiliary matrix Vehicle controller command: .in, Lyapunov matrix is ​​a symmetric positive definite matrix used to characterize the bounded set of error states on an ellipsoid.

[0088] To ensure that the error of a real vehicle relative to its nominal trajectory remains bounded under external disturbances and parameter mismatches, the vehicle controller constructs an error robust positive invariant set offline. In one implementation, if the discrete generalized perturbation The following constraints must be met: (32); in, For discrete generalized perturbation The known upper bound of the norm. Based on the expression (24) of the closed-loop error system, and combined with the auxiliary matrix. Lyapunov matrix And the upper bound of the discrete generalized disturbance, the vehicle controller for the error state vector Constructing error-robust positive invariant sets The expression for the error robust positive invariant set is: (33); in, j Let be the dimension of the error state vector; For error robust positive invariant set The boundary parameters of the ellipsoid are positive scalars; For the set of real numbers, express j A real vector space.

[0089] In this embodiment, the vehicle controller according to Based on performance indicators, upper bounds of discrete generalized perturbations, and decay margin parameters of the closed-loop error system, determine the robust positive invariant set of errors. The ellipsoidal boundary parameters are calculated using the following formula: (34); in, The decay margin parameter of the closed-loop error system is a scalar with a value greater than zero.

[0090] Step 702: Based on the error robust positive invariant set, calculate the control input robust tightening amount for each control channel and the state robust tightening amount for each constrained state component.

[0091] It should be noted that, in obtaining the error-robust positive invariant set... Subsequently, the vehicle controller further calculates the control input tightening amount offline. For the first... n One control channel, set Feedback gain matrix Then If so, then the first The formula for calculating the robust tightening amount of the control input for each control channel is: (35); in, For the first Robust tightening amount of control input for each control channel; This is the sequence number of the control channel. The first of the feedback gain matrix OK; For the feedback gain matrix, the first Transpose of a line.

[0092] The vehicle controller further calculates the state tightening amount offline. For the first... Let there be a constrained state component, and let... For use in extracting the first m The row vector of the nth state component, then the nth m The formula for calculating the state robust tightening amount of a constrained state component is: (36); in, For the first The state robust tightening of a constrained state component; This represents the index of the constrained state component. For use in extracting the first Row vectors of state components; row vector The transpose of .

[0093] Step 703: Based on the robust tightening amount of the control input for each control channel and each constrained state component, construct the control input tightening set and the state tightening set.

[0094] Step 704: Based on the control input tightening set and the state tightening set, perform tightening calculations on the original control input constraint set and the original state constraint set to obtain the tightened control input constraint set and the tightened state constraint set.

[0095] The formulas for calculating the tightened set of control input constraints and the tightened set of state constraints are as follows: (37); in, For the original set of control input constraints, For the original set of state constraints, This is the tightened set of control input constraints. This is the tightened set of state constraints. This represents the difference operation between Pontryagin and Pontryagin. Tighten the set to control the input; For a compacted set of states.

[0096] It should be noted that through the above constraint set tightening calculation, the vehicle controller pre-deducts the error offset that the real vehicle may produce under worst-case parameter mismatch and external disturbances, thereby ensuring that the upper-level nominal model predictive control is planned only within a more stringent virtual safety boundary. As long as the upper-level nominal model predictive control has a solution within the tightened constraint set, the real vehicle will still not violate the original physical constraints after superimposing error feedback compensation and worst-case discrete generalized disturbances.

[0097] In another exemplary embodiment of this application, based on the discrete-time nominal recursive equation, the tightened set of control input constraints, and the tightened set of state constraints, the vehicle controller constructs an upper-level nominal model predictive control optimization problem. In this case, step 206 is replaced by steps 801-804: Step 801: Define the nominal control increment and set the prediction time domain length and control time domain length.

[0098] It should be noted that at each sampling time The vehicle controller uses the current discrete nominal lateral state vector As initial values ​​for optimization, the prediction time domain and length control time domain length are set, and the future nominal control sequence is used as the optimization variable. Among these, To predict the length of the time domain, To control the length of the time domain.

[0099] The vehicle controller defines a nominal control increment. The expression for the nominal control increment is: (38); in, This represents the change in the nominal control quantity between future adjacent sampling times; For the vehicle at future sampling time The nominal control quantity; For the vehicle at future sampling time The nominal control quantity.

[0100] Step 802: Based on the discrete nominal transverse state vector, the nominal control increment, the nominal control increment weight matrix, the preset state weight matrix, and the control weight matrix in the discrete-time nominal recursive equation, construct the upper-level nominal model predictive control cost function.

[0101] It should be noted that the expression for the predictive control cost function of the upper-level nominal model constructed by the vehicle controller is as follows: (39); in, Predict the control cost function for the upper-level nominal model; To predict the step size index in the time domain; For the vehicle at future sampling time The nominal lateral state vector; For the vehicle at future sampling time The reference lateral state vector; For the vehicle at future sampling time The nominal control quantity; For the vehicle at future sampling time The nominal control increment satisfies equation (38); symbol Represents the transpose of a vector or matrix; This is the nominal control increment weight matrix. The nominal control increment weight matrix is ​​used to suppress drastic changes in the nominal control quantity between adjacent sampling times, thereby improving vehicle steering smoothness.

[0102] In order to make the planning objectives of the upper nominal model predictive control consistent with those of the lower level Robust feedback control maintains consistency in its disturbance rejection objectives. The lower-level performance output construction and the upper-level nominal model predictive control cost function share the same set of state weight matrices and control weight matrices, thereby achieving cost alignment between the upper and lower-level control objectives and avoiding conflicting control behaviors between the upper-level trajectory planning and the lower-level feedback compensation.

[0103] Step 803: Based on the tightened set of control input constraints, the tightened set of state constraints, the discrete-time nominal recursive equation, and the nominal control increment constraint set, determine the constraints of the upper-level nominal model predictive control.

[0104] The expression for the constraint conditions of the upper-level nominal model predictive control is: (40); in, For the vehicle at future sampling time The discrete reference path curvature; This is the set of nominal control incremental constraints.

[0105] Step 804: Determine the nominal control quantity generated by the upper nominal model predictive control plan based on the prediction time domain length, the control time domain length, the constraints of the upper nominal model predictive control, and the upper nominal model predictive control cost function.

[0106] It should be noted that at each sampling time The vehicle controller solves the nominal model predictive control optimization problem as described below online.

[0107] The expression for the nominal model predictive control optimization problem is: (41); in, For the vehicle at future sampling time The nominal control quantity.

[0108] After solving the nominal model predictive control optimization problem, the vehicle controller obtains the current sampling time. nominal control quantity .

[0109] By constructing the aforementioned nominal model predictive control optimization problem, the upper-level nominal model predictive control can not only achieve reference trajectory tracking within the tightened safety constraints, but also, through interaction with the lower-level... Robust feedback control uses a shared state weight matrix and control weight matrix to anticipate the trade-off between the lower-level disturbance target and the control effect, thereby reducing target conflict between upper and lower level controls and improving steering smoothness and control stability in high-speed, high-interference scenarios.

[0110] In another exemplary embodiment of this application, the hierarchical robust model predictive control process is further included in the online execution, i.e., after the discrete solution of the lower layer is completed... After the nominal control quantity generated by the offline upper-level nominal model predictive control programming based on the tightened set of control input constraints and the tightened set of state constraints with robust feedback gain, the vehicle controller performs the following steps at each discrete sampling sequence: Corresponding sampling time The following online control sub-steps are also executed.

[0111] Obtain vehicle at sampling time The real discrete lateral state vector It also obtains or calculates the reference path information in the current sampling time and the prediction time domain.

[0112] Acquire or read vehicle data maintained internally by the vehicle controller at the sampling time. Discrete nominal transverse state vector .

[0113] The vehicle's position at the sampling time is calculated according to equation (19). Error state vector .

[0114] Based on the constructed upper-level nominal model predictive control optimization problem, under the constraint shown in equation (40), the vehicle's performance at the sampling time can be obtained online. nominal control quantity .

[0115] Based on the lower-level robust feedback gain matrix obtained offline. And calculate the vehicle at the sampling time according to formula (23). Feedback compensation amount .

[0116] According to formula (16), the nominal control quantity With feedback compensation amount Synthesized as vehicles at the sampling time The total control quantity actually output to the steering actuator The total control quantity is then output to the steering actuator.

[0117] The discrete nominal lateral state vector is recursively updated according to equation (17) to obtain the vehicle's state at the next sampling time. Discrete nominal transverse state vector and will Store as the nominal initial state for the next control cycle.

[0118] The reference path information includes at least the current sampling time. and the reference path curvature sequence in the prediction time domain It may further include the corresponding reference lateral state sequence. .in, For vehicles at the sampling time The discrete reference path curvature; For the vehicle in the Each sampling time The reference lateral state vector.

[0119] Through the above online execution process, the vehicle controller uses upper-level nominal model predictive control to achieve forward-looking optimized tracking of the reference trajectory, and uses lower-level robust feedback control to compensate for the error state between the real vehicle and the nominal model in real time. Thus, even in the presence of parameter mismatch, unmodeled dynamics, and external physical disturbances, it can achieve high-precision, high-stability, and high-safety control of the lateral movement of commercial vehicles.

[0120] This application proposes a lateral hierarchical robust predictive control method for commercial vehicles, which breaks through the performance bottleneck of traditional control under extreme conditions. Its advantages include one or more of the following aspects.

[0121] Constructing a layered robust architecture adapted to harsh working conditions: The upper layer uses a nominal MPC to handle global optimal forward tracking and constraint processing, while the lower layer is based on an error system. The feedback is responsible for the local robust stability of the layered architecture, which is specifically designed for high-speed and high-interference scenarios.

[0122] Unified modeling of complex unmodeled dynamics of generalized disturbances: Unmodeled dynamics such as parameter mismatch, tire nonlinearity, and actuator hysteresis, which are difficult to obtain accurately when commercial vehicles are driving at high speeds, are represented by a unified set of external physical disturbances such as strong crosswinds as "discrete generalized disturbances", thereby constructing an error dynamic system that can fully cover extreme working conditions.

[0123] Theoretical stability guarantee based on linear matrix inequality (LMI): By solving the lower-level robust feedback gain offline using LMI, the lower-level controller can achieve strict bounded suppression of discrete generalized disturbances containing large modeling errors, thus theoretically guaranteeing the robust stability of the underlying error system.

[0124] Recursive feasibility guarantee based on robust positive invariant sets: The tightened boundaries of control variables and states are obtained offline analytically based on the robust positive invariant sets of the error system under worst-case perturbations. This mechanism ensures that as long as the upper-level nominal MPC has a solution within the tightened boundaries, the real vehicle will never exceed the boundary when subjected to strong disturbances, thus guaranteeing the recursive feasibility of the system without increasing the online computational load.

[0125] A cost-alignment mechanism for eliminating control conflicts: innovatively placing the lower layer The robust performance weighted matrix of the control is mapped and fused into the cost function of the upper-level nominal MPC. This mechanism enables the upper-level optimization direction to perceive in advance and actively adapt to the robust compensation trend of the lower level, mathematically eliminating the inconsistency between the upper and lower level control objectives, effectively avoiding high-frequency jitter of steering commands under high-speed and strong interference, and improving control smoothness.

[0126] This application defines discrete generalized disturbances and employs a lower-level robust feedback control compensation mechanism. Under high-speed, strong disturbances, the method described in this application achieves a more significant stability improvement compared to traditional single-layer MPC. Traditional single-layer MPC relies heavily on accurate vehicle dynamics models. When commercial vehicles face inaccurate dynamics model parameters under different loads and severe operating conditions such as strong crosswinds, it is highly susceptible to control divergence or vehicle instability due to severe mismatch in the dynamics model. The method in this application, through a layered architecture, mathematically encapsulates all parameter distortions, actuator hysteresis, and external strong disturbances into discrete generalized disturbances, and utilizes a lower-level robust feedback control compensation mechanism. The controller performs indiscriminate suppression. This mechanism of "lumped equivalence and underlying support" enables the system to maintain extremely high lateral stability even under extreme physical distortions.

[0127] This application presents a method for constructing a robust positive invariant set of errors and tightening the control input constraint set and state constraint set. Compared with traditional robust control methods, this method has strict physical safeguards and ensures the feasibility of the solution. Traditional robust control (such as traditional Tube MPC) often suffers from computational timeouts due to solving complex min-max optimization problems online, or from unsolvable online solutions due to overly conservative constraint handling. This method constructs a robust positive invariant set of errors offline and uses it to pre-deduct the limit safety margin to tighten the constraint boundary. This allows the online nominal MPC to solve a standard quadratic programming problem only within a more stringent "virtual boundary." As long as the upper-level programming has a solution, the real vehicle will never cross the lane line or experience steering saturation even after the worst-case physical disturbances are superimposed, thus ensuring both strict "recursive feasibility" and boundary-avoidance safety while maintaining real-time performance.

[0128] Furthermore, this application achieves cost alignment by constructing an upper-level nominal model to predict the control cost function, resulting in superior control smoothness by eliminating internal friction and game-theoretic conflicts compared to traditional hierarchical control. In traditional hierarchical control architectures, upper and lower level controllers often operate independently. The upper level pursues trajectory smoothness, while the lower level frequently outputs drastic reverse correction commands to maintain a safety baseline. This internal struggle between the two at the execution end easily leads to "steering wheel vibration" during high-speed driving. This method innovatively introduces a cost alignment mechanism, physically "embedding" the robust performance weights of the lower level into the cost function of the upper-level MPC. This allows the upper level to "foresee and understand" the disturbance rejection pressure of the lower level during planning, actively avoiding nominal trajectories that would induce drastic corrections at the lower level. This mechanism eliminates the internal friction and conflict of controllers at the root, significantly improving the ride comfort and steering smoothness of commercial vehicles at high speeds.

[0129] This application achieves a balance between robustness and the real-time performance required for automotive-grade chips by offline calculation of feedback gain and offline tightening of control input constraint sets and state constraint sets. This application transforms the complex calculation of robust feedback gain into offline computation of linear matrix inequalities, while also moving the tightening process of state variables and control input variables to the offline stage. In the online control stage, the vehicle controller only needs to call the offline-calculated tightening constants and feedback gain for simple algebraic operations and standard MPC solutions, significantly reducing online computing power consumption and fully meeting the millisecond-level real-time response requirements of existing automotive-grade control chips for commercial vehicles.

[0130] Based on the same inventive concept, this application also provides a device for implementing the aforementioned lateral hierarchical robust predictive control for commercial vehicles. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the lateral hierarchical robust predictive control for commercial vehicles provided below can be found in the limitations of the lateral hierarchical robust predictive control method for commercial vehicles described above, and will not be repeated here.

[0131] In one exemplary embodiment, a lateral hierarchical robust predictive control device for commercial vehicles is provided, including a vehicle control mechanism and a steering actuator.

[0132] The vehicle control mechanism is used to execute the commercial vehicle lateral hierarchical robust predictive control method described in any of the above embodiments and output the total control quantity.

[0133] The steering actuator is connected to the vehicle control mechanism and is used to receive the total control quantity and perform steering actions according to the total control quantity.

[0134] In one exemplary embodiment, a computer device is provided, which may be a server or a terminal, and its internal structure diagram may be as follows. Figure 3 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores basic vehicle information. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When executed by the processor, the computer program implements a lateral hierarchical robust predictive control method for commercial vehicles.

[0135] Figure 3 The structures shown are merely block diagrams of some structures related to the present application and do not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than shown in the figures, or combine certain components, or have different component arrangements. In an exemplary embodiment, a computer device is provided, including a memory and a processor. The memory stores a computer program, and the processor executes the computer program to implement the steps in the above-described method embodiments.

[0136] In one exemplary embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0137] In one exemplary embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above-described method embodiments.

[0138] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of related data must comply with relevant regulations and be authorized by the owner of the corresponding device.

[0139] Those skilled in the art will understand that all or part of the processes in the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments described above. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM).

[0140] The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0141] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0142] This document uses specific examples to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the methods and core ideas of this application. Furthermore, those skilled in the art will recognize that, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A lateral hierarchical robust predictive control method for commercial vehicles, characterized in that, The lateral hierarchical robust predictive control method for commercial vehicles includes: Obtain basic vehicle information; the basic vehicle information includes sampling period, current longitudinal speed of the vehicle, reference path information, vehicle's true discrete lateral state vector, and nominal vehicle parameters; Based on the vehicle's basic information, a continuous-time dynamics model is constructed, and linearization, discretization, and nominal state recursion are performed on the continuous-time dynamics model to determine the discrete-time nominal recursive equation. Based on the discrete nominal lateral state vector in the discrete-time nominal recursive equation and the vehicle's real discrete lateral state vector, the error state update equation is determined. Based on the aforementioned error state update equation, the lower layer is discretely solved. Robust feedback gain is used to obtain the lower-level robust feedback gain matrix and the feedback compensation amount generated by the lower-level robust feedback control; Based on the lower-level robust feedback gain matrix and the error state update equation, an error robust positive invariant set is constructed, and the tightened control input constraint set and the tightened state constraint set are determined according to the error robust positive invariant set. Based on the discrete-time nominal recursive equation, the tightened set of control input constraints, and the tightened set of state constraints, the nominal control quantity generated by the upper-level nominal model predictive control program is determined. Based on the feedback compensation amount and the nominal control amount, the total control amount of the actual output is obtained; the total control amount is used as the input amount of the vehicle steering actuator, so that the steering actuator performs steering action according to the total control amount.

2. The lateral hierarchical robust predictive control method for commercial vehicles according to claim 1, characterized in that, Based on the vehicle's basic information, a continuous-time dynamics model is constructed. This model is then linearized, discretized, and subjected to nominal state recursion to determine the discrete-time nominal recursive equations. Specifically, this includes: Based on the two-dimensional monorail model and the vehicle's basic information, a continuous-time dynamics model is constructed. The continuous-time dynamics model is linearized to obtain the continuous-time nominal state-space equation; The continuous-time nominal state-space equations are discretized using the explicit Euler discretization method to obtain a discrete-time dynamic model. Based on the discrete-time dynamics model, a discrete-time nominal dynamics model is determined, and a nominal state recursion is performed based on the discrete-time nominal dynamics model to determine the discrete-time nominal recursion equation.

3. The lateral hierarchical robust predictive control method for commercial vehicles according to claim 1, characterized in that, The step of determining the error state update equation based on the discrete nominal lateral state vector in the discrete-time nominal recursive equation and the vehicle's actual discrete lateral state vector specifically includes: The difference between the discrete nominal lateral state vector in the discrete-time nominal recursive equation and the vehicle's real discrete lateral state vector is calculated to obtain the error state vector. Based on the error state vector, determine the error state vector at the next moment; The parameter mismatch term, the unmodeled dynamic influence term, and the external physical disturbance term in the error state vector at the next moment are defined as discrete generalized disturbances; Based on the generalized perturbation assignment matrix and the discrete generalized perturbation, the error state update equation is determined.

4. The lateral hierarchical robust predictive control method for commercial vehicles according to claim 3, characterized in that, The lower layer is discretely solved based on the error state update equation. Robust feedback gain, obtained by combining the feedback gain matrix with the feedback compensation amount generated by the lower-level robust feedback control, specifically includes: The lower-level robust feedback controller adopts a state feedback form, and the error closed-loop system is obtained based on the error state update equation. Preset the state weight matrix and control weight matrix, and construct the lower-level performance output; set up Performance indicators, determined The constraints; The expression for the constraint condition is: ； in, Let be the transfer function of the closed-loop error system from the discrete generalized perturbation to the lower-level performance output. This is a preset upper bound for robust performance; The constraints are transformed into linear matrix inequalities by using the discrete-time bounded real lemma and Lyapunov stability theory. The linear matrix inequality is solved offline using the interior point method to determine the lower-level robust feedback gain matrix and the feedback compensation amount generated by the lower-level robust feedback control.

5. The lateral hierarchical robust predictive control method for commercial vehicles according to claim 4, characterized in that, The process employs the discrete-time bounded real lemma and Lyapunov stability theory to transform the constraints into linear matrix inequalities, specifically including: By defining the Lyapunov matrix and a symmetric positive definite matrix, and introducing auxiliary variables, the constraints on the feedback gain matrix are obtained; the expression for the constraints on the feedback gain matrix is ​​as follows: ； in, K For the feedback gain matrix, Y For auxiliary matrix; L As an auxiliary variable; Y -1 It is the inverse of the auxiliary matrix; Combining the Schul complement lemma and congruent transformation, the aforementioned The constraints are transformed into linear matrix inequalities.

6. The lateral hierarchical robust predictive control method for commercial vehicles according to claim 1, characterized in that, The process of constructing an error robust positive invariant set based on the lower-level robust feedback gain matrix and the error state update equation, and determining the tightened control input constraint set and the tightened state constraint set based on the error robust positive invariant set, specifically includes: Based on the lower-level robust feedback gain matrix and the error state update equation, an error robust positive invariant set is constructed. Based on the aforementioned error robust positive invariant set, calculate the control input robust tightening amount for each control channel and the state robust tightening amount for each constrained state component; Based on the robust tightening amount of the control input for each control channel and each constrained state component, a control input tightening set and a state tightening set are constructed. Based on the tightened control input set and the tightened state set, and by performing tightening calculations on the original control input constraint set and the original state constraint set, the tightened control input constraint set and the tightened state constraint set are obtained.

7. The lateral hierarchical robust predictive control method for commercial vehicles according to claim 6, characterized in that, The calculation formula for the robust tightening amount of the control input of each control channel is as follows: ； in, For the first Robust tightening amount of control input for each control channel; This is the sequence number of the control channel. The first of the feedback gain matrix OK; For the feedback gain matrix, the first transpose of a line; For the ellipsoidal boundary parameters of the error-robust positive invariant set; The formula for calculating the state robust tightening amount of each constrained state component is as follows: ； in, For the first The state robust tightening of a constrained state component; This represents the index of the constrained state component. For use in extracting the first Row vectors of state components; row vector Transpose of; The formulas for calculating the tightened set of control input constraints and the tightened set of state constraints are as follows: ； in, For the original set of control input constraints, For the original set of state constraints, This is the tightened set of control input constraints. This is the tightened set of state constraints. This represents the difference operation between Pontryagin and Pontryagin. Tighten the set to control the input; For a compacted set of states.

8. The lateral hierarchical robust predictive control method for commercial vehicles according to claim 4, characterized in that, The determination of the nominal control quantity generated by the upper-level nominal model predictive control programming based on the discrete-time nominal recursive equation, the tightened set of control input constraints, and the tightened set of state constraints specifically includes: Define the nominal control increment and set the prediction time domain length and control time domain length; Based on the discrete nominal horizontal state vector, the nominal control increment, the nominal control increment weight matrix, the preset state weight matrix, and the control weight matrix in the discrete-time nominal recursive equation, the upper-level nominal model predicts the control cost function. Based on the tightened set of control input constraints, the tightened set of state constraints, the discrete-time nominal recursive equation, and the set of nominal control increment constraints, the constraints of the upper-level nominal model predictive control are determined. Based on the prediction time domain length, the control time domain length, the constraints of the upper-level nominal model predictive control, and the upper-level nominal model predictive control cost function, the nominal control quantity generated by the upper-level nominal model predictive control planning is determined.

9. A robust predictive control device for lateral hierarchical control of commercial vehicles, characterized in that, This includes vehicle control mechanisms and steering actuators; The vehicle control mechanism is used to execute the commercial vehicle lateral hierarchical robust predictive control method according to any one of claims 1-8, and output the total control quantity; The steering actuator is connected to the vehicle control mechanism and is used to receive the total control quantity and perform steering actions according to the total control quantity.

10. A computer product comprising a computer program, characterized in that, When executed by a processor, the computer program implements the lateral hierarchical robust predictive control method for commercial vehicles as described in any one of claims 1-8.

Images