Vehicle lateral-longitudinal-vertical force coupling model predictive control method, device and equipment

Abstract

This application relates to the field of vehicle control technology, specifically providing a method, apparatus, and device for model predictive control of vehicle lateral, longitudinal, and vertical force coupling. The method includes: obtaining initial weight values ​​for a model predictive controller based on a subjective and objective quantitative model of vehicle driving and an optimization objective; the subjective and objective quantitative model characterizes the mapping relationship between the driver's subjective evaluation data and the vehicle's motion state data; obtaining the vehicle's expected control quantity based on the initial weight values ​​and the model predictive controller; and optimizing the initial weight values ​​based on the vehicle's expected control quantity and the vehicle's current driving data to obtain an optimized model predictive controller; wherein the optimized model predictive controller is used to control the vehicle's output control quantity during driving to meet the driver's personalized driving needs.

Description

Technical Field

[0001] This application relates to the field of vehicle control technology, and more specifically, to a method, apparatus, and equipment for predictive control of vehicle lateral, longitudinal, and vertical force coupling model. Background Technology

[0002] With the development of intelligent and electric vehicles, vehicle dynamics control technology is increasingly becoming a key to improving active safety and driving performance.

[0003] Currently, independent control strategies are typically employed for the lateral (steering), longitudinal (acceleration / braking), and vertical (suspension) dynamics control of vehicles. Examples include anti-lock braking systems (ABS) and lateral stability control. However, due to the strong coupling nonlinear relationships between vehicle dynamics in the lateral, longitudinal, and vertical directions, independent control in one direction may lead to performance degradation in other directions. Existing coupled lateral, longitudinal, and vertical vehicle dynamics control methods are usually based on fixed-weight model predictive control methods. Their weight allocation heavily relies on experience and calibration, and the actual control effect may deviate from the expectations of real users and fail to meet the personalized needs of drivers.

[0004] Therefore, how to provide a technical solution for personalized vehicle lateral and vertical force coupling model predictive control has become a technical problem that needs to be solved. Summary of the Invention

[0005] The purpose of some embodiments of this application is to provide a method, apparatus, and device for predictive control of a vehicle's lateral, longitudinal, and vertical force coupling model. The technical solutions of the embodiments of this application can accurately control the output of the vehicle's lateral, longitudinal, and vertical force coupling prediction model to meet the user's personalized driving needs.

[0006] In a first aspect, some embodiments of this application provide a method for model predictive control of vehicle lateral, longitudinal, and vertical force coupling, comprising: obtaining initial weight values ​​of a model predictive controller based on a subjective and objective quantitative model of vehicle driving and an optimization objective; wherein the subjective and objective quantitative model characterizes the mapping relationship between the driver's subjective evaluation data and the vehicle's motion state data; the model predictive controller includes a target control function and weights for different vehicle operating states; obtaining a vehicle expected control quantity based on the initial weight values ​​and the model predictive controller; optimizing the initial weight values ​​based on the vehicle expected control quantity and the vehicle's current driving data to obtain an optimized model predictive controller; wherein the optimized model predictive controller is used to control the vehicle's output control quantity during driving.

[0007] Some embodiments of this application obtain the initial weight values ​​of the model predictive controller by first obtaining the expected control quantity of the vehicle through subjective and objective quantitative models and optimization objectives. The expected control quantity of the vehicle and the current driving data of the vehicle are analyzed and optimized to obtain the optimized model predictive controller. This achieves precise optimization of the model predictive controller, accurately controls the output of the controller, improves the handling stability of the vehicle, and meets the personalized driving needs of users.

[0008] In some embodiments, obtaining the initial weight value of the model predictive controller based on the subjective and objective quantitative model of vehicle driving and the optimization objective includes: solving the subjective and objective quantitative model with the preset score value corresponding to the vehicle model as the optimization objective to obtain the initial weight value.

[0009] Some embodiments of this application use the vehicle model and preset score as optimization objectives to solve the subjective and objective quantitative model and obtain the initial weight value, thus overcoming the dependence of the weight parameters on expert experience and using different vehicle models and platforms.

[0010] In some embodiments, obtaining the expected control quantity of the vehicle based on the initial weight value and the model predictive controller includes: predicting the expected trajectory of multiple parameters based on the throttle opening signal and steering wheel rotation signal of the vehicle; wherein the multiple parameters include the longitudinal speed, lateral speed, vertical speed, yaw rate, pitch rate and roll rate of the vehicle; obtaining multiple control quantities based on the expected trajectory and a portion of the weights in the initial weight value; and allocating the multiple control quantities using the weights other than the portion of the weights in the initial weight value and the model predictive controller to determine the expected control quantity of the vehicle.

[0011] Some embodiments of this application first predict the desired trajectory using the vehicle's throttle opening signal and steering wheel rotation signal, then combine the initial weight values ​​to obtain multiple control quantities, and then allocate the multiple control quantities to obtain the vehicle's expected control quantity, thereby achieving the effective output of the vehicle's expected control quantity.

[0012] In some embodiments, obtaining multiple control quantities based on the desired trajectory and a portion of the initial weight values ​​includes: constructing the vehicle dynamics equations of the vehicle; constructing a control loss function based on the vehicle dynamics equations; wherein the control loss function includes the error loss between the desired trajectory and the current predicted trajectory and the control quantity transformation loss; obtaining the optimal control quantity output by the control loss function by minimizing the loss and the portion of the weights; wherein the optimal control quantity serves as the multiple control quantities.

[0013] Some embodiments of this application determine the optimal control quantity by constructing vehicle dynamics equations and control loss functions, thereby achieving optimized allocation of the control quantity and improving overall performance.

[0014] In some embodiments, the step of allocating the plurality of control quantities using the weights other than the partial weights in the initial weight values ​​and the model predictive controller to determine the expected control quantity of the vehicle includes: constructing a force distribution loss function related to tire load rate; solving for the target value of the force distribution loss function under minimum loss using the other weights; and obtaining the expected control quantity of the vehicle using the target value and the force distribution control dynamics model in the model predictive controller.

[0015] Some embodiments of this application determine the target value through a force distribution loss function and other weights, and then combine it with a force distribution control dynamics model to obtain the expected control quantity of the vehicle, thereby achieving optimized distribution of lateral, longitudinal, and vertical cooperative forces and achieving optimal global performance.

[0016] In some embodiments, optimizing the initial weight value based on the expected vehicle control quantity and the current vehicle driving data to obtain an optimized model predictive controller includes: acquiring vehicle control index features corresponding to the current vehicle driving data; adjusting the initial weight value by calculating the error value between the expected vehicle control quantity and the vehicle control index features, as well as user behavior, to obtain the optimized model predictive controller.

[0017] Some embodiments of this application optimize the initial weight values ​​by using the error values ​​between vehicle control index characteristics and the expected control quantities of the vehicle and user behavior to obtain an optimized model predictive controller, thereby achieving the optimal global performance of the vehicle under complex driving conditions, and thus improving the vehicle's handling stability, safety and ride comfort under extreme conditions.

[0018] In some embodiments, the subjective-objective quantification model is constructed through the following steps: obtaining time series data corresponding to objective indicators of the vehicle under different operating conditions, and obtaining subjective scores under the different operating conditions; processing the time series data to obtain feature values ​​of the objective indicators; and obtaining the subjective-objective quantification model based on the mapping relationship between the feature values ​​of all objective indicators and the subjective scores.

[0019] Some embodiments of this application construct subjective-objective quantitative models using subjective-objective related data, overcoming the dependence of weight parameters on expert experience, enabling the control algorithm to adapt to different vehicle platforms, and significantly reducing calibration costs and development cycles.

[0020] In some embodiments, the method further includes: acquiring user feedback behavior in response to in-vehicle prompts and monitoring user intervention behavior during driving; and correcting the subjective-objective quantification model based on the feedback behavior and the intervention behavior.

[0021] Some embodiments of this application improve the accuracy and robustness of subjective and objective quantitative models by correcting them through feedback and intervention behaviors.

[0022] Secondly, some embodiments of this application provide a device for model predictive control of vehicle lateral, longitudinal, and vertical force coupling, comprising: a weight acquisition module, used to acquire initial weight values ​​of a model predictive controller based on a subjective and objective quantitative model of vehicle driving and an optimization objective; wherein the subjective and objective quantitative model represents the mapping relationship between the driver's subjective evaluation data and the vehicle's motion state data; the model predictive controller includes a target control function and weights for different vehicle operating states; an expected control module, used to acquire the expected control quantity of the vehicle based on the initial weight values ​​and the model predictive controller; and an optimization control module, used to optimize the initial weight values ​​based on the expected control quantity of the vehicle and the vehicle's current driving data to obtain an optimized model predictive controller; wherein the optimized model predictive controller is used to control the output control quantity of the vehicle during driving.

[0023] Thirdly, some embodiments of this application provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can implement the method described in any embodiment of the first aspect.

[0024] Fourthly, some embodiments of this application provide an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor, when executing the program, can implement the method as described in any embodiment of the first aspect.

[0025] Fifthly, some embodiments of this application provide a computer program product, the computer program product including a computer program, wherein the computer program, when executed by a processor, can implement the method described in any embodiment of the first aspect. Attached Figure Description

[0026] To more clearly illustrate the technical solutions of some embodiments of this application, the accompanying drawings used in some embodiments of this application will be briefly described below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0027] Figure 1 A flowchart of a method for predictive control of vehicle lateral, longitudinal, and vertical force coupling model provided for some embodiments of this application;

[0028] Figure 2One of the simulation diagrams of an optimized model predictive controller provided for some embodiments of this application;

[0029] Figure 3 A second simulation diagram of an optimized model prediction controller provided for some embodiments of this application;

[0030] Figure 4 Block diagram of a vehicle lateral, longitudinal and vertical force coupling model predictive control device provided for some embodiments of this application;

[0031] Figure 5 A schematic diagram of an electronic device provided for some embodiments of this application. Detailed Implementation

[0032] The technical solutions of some embodiments of this application will now be described with reference to the accompanying drawings.

[0033] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0034] In related technologies, studies considering lateral, longitudinal, and vertical coupling control often employ model predictive control (MPC) methods, balancing control performance in different directions through weight allocation in the optimization function. In MPC, the cost function of the optimization problem is typically a weighted sum of multiple sub-objectives, such as tracking accuracy, ride comfort, and handling stability; the weight allocation directly determines the controller's trade-offs across different performance dimensions. Current weight settings largely rely on the prior knowledge of control engineers and extensive simulation debugging, followed by subjective evaluation through real-vehicle testing under common operating conditions. This results in a massive calibration workload and poor generalization ability. In other words, a set of carefully calibrated weight parameters for a specific vehicle model and configuration is not applicable to other models, severely limiting the universality and scalability of the control strategy. Furthermore, controller design can only use objective indicators as optimization targets, but the final evaluation of the control strategy's effectiveness is usually based on subjective feelings during performance evaluation. Currently, there is a lack of a widely accepted mathematical model that can accurately quantify human subjective feelings. This leads to a lack of trade-off criteria when multiple control objectives conflict, forcing engineers to guess and set trade-off criteria based on personal driving experience. This could cause the controller's driving style to deviate from the expectations of real users and make it difficult to meet the personalized needs of drivers.

[0035] Existing lateral, longitudinal, and vertical coupled vehicle dynamics control methods are typically based on fixed-weight model predictive control. Their weight allocation heavily relies on experience and calibration, requiring subjective evaluation and calibration through real-vehicle testing under different operating conditions. This limits the versatility and scalability of the control strategy. Furthermore, there is currently a lack of a model mapping subjective experience to objective indicators. Even with subjective evaluation and weight calibration, the actual controller performance may deviate from real user expectations and fail to meet the driver's personalized needs.

[0036] As can be seen from the above-mentioned related technologies, in the existing technologies, due to the highly nonlinear and time-varying nature of the lateral, longitudinal, and vertical dynamic coupling relationship, it is difficult to analytically determine the optimal weights; there is a lack of methods that can perform online adaptive optimization based on real-time vehicle status and driving intentions; the weight allocation of the controller mainly relies on experience and calibration, and the current weight allocation method is highly subjective, inefficient, and costly, and cannot adapt to changing environments and vehicle states, which seriously restricts the practical application effect of model predictive control in lateral, longitudinal, and vertical coupling control; there is a lack of correspondence between subjective perception and objective indicators of driving performance, and it is difficult to determine the trade-off criteria when multiple objectives conflict, so the weight allocation can only be set based on subjective perception, and the actual effect of the controller may deviate from the expectations of real users and is difficult to meet the personalized needs of drivers.

[0037] In view of this, some embodiments of this application provide a method for model predictive control of vehicle lateral, longitudinal, and vertical force coupling. In this method, the initial weight values ​​of the model predictive controller can be obtained by subjectively and objectively quantifying the model and the optimization objective. Then, the expected vehicle control quantity output by the model predictive controller is obtained based on the initial weight values. Finally, the model predictive controller is optimized according to the expected vehicle control quantity and the current vehicle driving data to obtain the optimized model predictive controller. The optimized model predictive controller can control the vehicle control quantity output during vehicle driving, thereby achieving the control effect of the controller and meeting the personalized needs of the driver.

[0038] The following is in conjunction with the appendix Figure 1 The present application provides an exemplary embodiment of the implementation process of vehicle lateral, longitudinal, and vertical force coupling model predictive control executed by a controller.

[0039] Please see the appendix Figure 1 , Figure 1 A flowchart illustrating a method for vehicle lateral, longitudinal, and vertical force coupling model predictive control is provided for some embodiments of this application. This method for vehicle lateral, longitudinal, and vertical force coupling model predictive control may include:

[0040] S110, Based on the subjective and objective quantitative model of vehicle driving and the optimization objective, obtain the initial weight values ​​of the model predictive controller; wherein, the subjective and objective quantitative model represents the mapping relationship between the driver's subjective evaluation data and the vehicle motion state data; the model predictive controller includes the objective control function and the weights under different vehicle operating states.

[0041] For example, in a specific embodiment of this application, the weights of the model prediction controller are initialized using the constructed subjective and objective quantification model and the pre-set optimization objective to obtain initial weight values.

[0042] In some embodiments of this application, the subjective-objective quantification model is constructed by: obtaining time series data corresponding to objective indicators of the vehicle under different operating conditions, and obtaining subjective scores under the different operating conditions; processing the time series data to obtain feature values ​​of the objective indicators; and obtaining the subjective-objective quantification model based on the mapping relationship between the feature values ​​of all objective indicators and the subjective scores.

[0043] For example, in a specific embodiment of this application, subjective evaluation data of the driver's driving comfort and / or drivability under various test conditions (i.e. different conditions) are obtained (as a specific example of subjective scoring), and motion state data of various objective indicators of the vehicle are collected simultaneously (as a specific example of time series data) to establish a subjective and objective quantitative model.

[0044] Specifically, typical operating condition testing and data acquisition involve subjective evaluation and objective index collection in experimental scenarios that can stimulate lateral, longitudinal, and vertical coupling and differentiate between comfort and handling characteristics. Objective indices can include step steering, cornering braking, acceleration out of a corner, and cornering bumps. Time-series data e(t) of the above objective indices and their derivatives are collected at high frequencies (e.g., 50-100Hz). Immediately after each operating condition, professional performance evaluators or test subjects provide subjective scores via an onboard terminal. The experiment is repeated under different control strategies by changing the parameters in the model predictive controller. The lengthy time-series data e(t) is extracted into feature values ​​highly correlated with human perception, including the root mean square value of the error and the frequency-weighted mean square value. Each operating condition is treated as a data sample. The input feature vector of the constructed objective quantitative model is composed of the feature values ​​of all objective indices, and the output is the average subjective score YC for that operating condition.

[0045] In some embodiments of this application, S110 may include: using a preset score value corresponding to the vehicle model as the optimization objective, solving the subjective and objective quantification model to obtain the initial weight value.

[0046] For example, in a specific embodiment of this application, based on the vehicle model positioning, and using the output score of the subjective and objective quantitative model (as a specific example of a preset score value) as the optimization objective, the initial weight values ​​corresponding to each motion state parameter (i.e., objective index) in the vehicle's control behavior are solved. Alternatively, a set of optimal weights (i.e., initial weight values) is solved such that the predicted subjective score corresponding to the control behavior generated by the Model Predictive Control (MPC, or MPC controller for short) is the highest under multiple test conditions. The vehicle models include SUVs, sedans, off-road vehicles, trucks, etc.; or, the vehicle models include family cars, commercial cars, etc.

[0047] S120, based on the initial weight values ​​and the model prediction controller, obtain the expected control quantity of the vehicle.

[0048] For example, in a specific embodiment of this application, a model predictive controller is constructed based on a vehicle lateral-longitudinal-vertical coupled dynamics model. The cost function of this model predictive controller includes multiple weight coefficients related to the lateral, longitudinal, and vertical motion states. Using the initial weight values ​​related to the lateral, longitudinal, and vertical motion states obtained above, the output of the vehicle lateral-longitudinal-vertical coupled dynamics model can be controlled to obtain the vehicle's expected control quantity. The vehicle's expected control quantity may include hub motor torque, front and rear wheel steering angles, and active suspension force. In addition to the initial weight values, the vehicle lateral-longitudinal-vertical coupled dynamics model also requires the input of user-inputted throttle opening and steering wheel rotation signals.

[0049] In some embodiments of this application, the MPC controller has a hierarchical structure, including a vehicle trajectory prediction layer, a force optimization layer, and an actuator control layer. The vehicle trajectory prediction layer predicts the desired six-degree-of-freedom (6DOF) motion trajectory of the vehicle, including the vehicle's velocities in the longitudinal, lateral, and vertical directions, as well as the angular velocities of pitch, roll, and yaw. The vehicle force optimization layer constructs the lateral, longitudinal, and vertically coupled six-DOF dynamic equations of the vehicle. Based on the desired six-DOF motion trajectory, model predictive control is used to optimize and calculate the trajectories corresponding to the twelve forces acting on the four tires in the lateral, longitudinal, and vertical directions. The actuator control layer constructs a dynamic model including hub motor torque (i.e., the four-wheel motor torque hereinafter referred to as four-wheel motor torque), front and rear wheel steering angles, and active suspension forces. Based on the model predictive control method, the actuator control quantities corresponding to the desired force trajectories obtained in the previous optimization step are calculated.

[0050] The implementation process of S120 is illustrated below.

[0051] In some embodiments of this application, S120 may include:

[0052] S121, based on the throttle opening signal and steering wheel rotation signal of the vehicle, predict the desired trajectory of multiple parameters; wherein, the multiple parameters include the longitudinal speed, lateral speed, vertical speed, yaw rate, pitch rate and roll rate of the vehicle.

[0053] For example, in some embodiments of this application, the vehicle trajectory prediction layer predicts the desired six-degree-of-freedom vehicle motion trajectory based on the user-input throttle opening signal and steering wheel rotation signal. This desired trajectory serves as the reference trajectory for the initial model predictive control. The specific implementation process is as follows:

[0054] The desired longitudinal velocity in the desired longitudinal velocity trajectory is determined by the initial velocity and the acceleration curve. The acceleration curve can be obtained by mapping the curve of the throttle opening signal input by the user, and the mapping relationship is adjusted according to different vehicle motion modes (such as comfort, sport, etc.). The formula for calculating the longitudinal velocity is:

[0055]

[0056] in, V x Indicates the desired longitudinal velocity. V x0 Indicates the initial longitudinal velocity. The acceleration curve mapped by the curve representing the throttle opening signal; Let t represent the desired angular velocity of the roll, and t represent time t. For time, d This indicates that calculus is performed over time.

[0057] The desired yaw rate in the desired trajectory is obtained from a linear two-degree-of-freedom reference model. Ignoring the transient response process, the calculation formula is:

[0058]

[0059] in, For the front wheel steering angle, l f and l r These represent the distances from the front and rear wheels to the center of mass, respectively. K g It is an understeering gradient. Let be the first derivative of the desired yaw rate.

[0060] The desired trajectory for the remaining degrees of freedom is determined by the fact that the four-wheel independent drive vehicle addressed in this application can easily achieve effective control over its lateral dynamic response and prevent lateral movement; therefore, the lateral reference speed is used. V y The trajectory is set to 0. The vertical velocity...V z The desired trajectory is also set to 0 to ensure vehicle ride comfort. In addition, to suppress vehicle pitch and roll movements, both the desired pitch angular velocity trajectory and the desired roll angular velocity trajectory are set to 0.

[0061] S122, Based on the desired trajectory and a portion of the weights in the initial weight values, obtain multiple control quantities.

[0062] For example, in a specific embodiment of this application, the function of the vehicle force optimization layer is to transform the reference desired trajectory generated by the vehicle trajectory prediction layer into a desired force trajectory.

[0063] In some embodiments of this application, S122 may include: constructing the vehicle dynamics equations of the vehicle; constructing a control loss function based on the vehicle dynamics equations; wherein the control loss function includes the error loss between the desired trajectory and the current predicted trajectory and the control quantity transformation loss; obtaining the optimal control quantity output by the control loss function by minimizing the loss and the partial weights; wherein the optimal control quantity is used as the plurality of control quantities.

[0064] For example, in a specific embodiment of this application, a six-degree-of-freedom vehicle dynamics equation that reflects the vehicle's longitudinal, lateral, and vertical motion is constructed. An objective function (as a specific example of a control loss function) is designed, and the desired six control quantities (as a specific example of multiple control quantities) are ultimately obtained through online optimization of the model predictive controller. The specific implementation process is as follows:

[0065] Based on the vehicle's physical model, the six-degree-of-freedom vehicle dynamics equations are constructed as follows:

[0066]

[0067] in, m For the overall vehicle quality, m s It is the sprung mass. g It is the acceleration due to gravity. F xij and F yij These represent the longitudinal and lateral forces on each wheel. F zij,s Let be the vertical force of each suspension on the sprung mass; ij∈{fl, fr, rl, rr}, which represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. I x and I y These are the moments of inertia for vehicle roll and pitch, respectively. Iz The yaw moment of inertia of the vehicle, t f and t r These are the tire distances on the front and rear axles, respectively. ρ For the desired pitch angular velocity, The first derivative of the desired pitch angular velocity. The second derivative of the desired pitch angular velocity; For the desired angular velocity of the roll, The first derivative of the desired angular velocity of the roll. This is the second derivative of the desired angular velocity of the roll. The first derivative of the desired yaw rate. Let be the second derivative of the desired yaw rate. l f and l r These represent the distances from the front and rear wheels to the center of mass, respectively. V x Indicates the desired longitudinal velocity. Characterization V x The derivative; V y For lateral reference speed, Characterization V y The derivative; V z Vertical velocity, Characterization V z The derivative of . All the above force vectors are in the vehicle body coordinate system.

[0068] Based on the six-degree-of-freedom vehicle dynamics equations, the six-degree-of-freedom differential equations of the vehicle can be expressed as follows:

[0069]

[0070] The specific meanings of the parameters within this formula can be found in the previous paragraph.

[0071] The loss function guides the optimization direction of the model predictive controller, obtaining the optimal control input by minimizing the loss. The loss function consists of two parts: the first part measures the difference between the predicted trajectory and the desired trajectory (i.e., error loss), and the second part controls the rate of change of lateral, longitudinal, and vertical forces (i.e., control input transformation loss), avoiding vehicle pitch and roll caused by frequent transformations. The first part is the trajectory error loss function. J The calculation formula is as follows:

[0072]

[0073] in, V x ( k+i | k ), V y ( k+i | k ), V z ( k+i | k ), ( k+i | k ), ( k+i | k ), ( k+i | k The numbers ) represent the vehicle's longitudinal, lateral, and vertical velocities, and its roll, pitch, and yaw angular velocities at time k+i, predicted using the current control input at time k, respectively. p represents the total number of time points. The target value for the desired longitudinal velocity, The target value for the desired yaw rate; Q 1~ Q 6 represents the weight (i.e., partial weight) of each state variable error in the total loss function.

[0074] The second part is the control input conversion rate loss. J 2 (i.e., control variable transformation loss) is calculated using the following formula:

[0075]

[0076] in, , , These represent the changes in force in the x, y, and z directions, respectively. , , These represent the changes in torque along the x, y, and z axes, respectively.

[0077] The overall loss function is then min J = J 1+ J 2.

[0078] By inputting partial weights J In 1, then with J To minimize the loss, the optimal control quantities of the loss function are obtained, namely, six control quantities: longitudinal, lateral, vertical force, pitch, roll, and yaw moments.

[0079] S123, using the weights other than the partial weights in the initial weight values ​​and the model prediction controller, the multiple control quantities are allocated to determine the expected control quantity of the vehicle.

[0080] For example, in a specific embodiment of this application, the vehicle type is a vehicle equipped with four-wheel independent drive, front and rear wheel steering, and active suspension. Therefore, a force distribution control dynamics model including the torque of the four-wheel motors, the steering angle of the front and rear wheels, and the active suspension force is constructed at the actuator control layer. Based on the model predictive control method, the actuator control quantity corresponding to the desired force trajectory (as a specific example of the expected control quantity of the vehicle) is calculated.

[0081] In some embodiments of this application, S123 may include: constructing a force distribution loss function related to tire load rate; solving for the target value of the force distribution loss function under minimum loss using the other weights; and obtaining the expected control quantity of the vehicle using the target value and the force distribution control dynamics model in the model predictive controller.

[0082] For example, in a specific embodiment of this application, in the force distribution control dynamics model, the longitudinal force in the tire coordinate system is directly determined by the torque. Ignoring the dynamics of wheel angular acceleration, the formula for calculating the wheel longitudinal force is as follows:

[0083]

[0084] Where ij∈{fl, fr, rl, rr}, representing the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively; F xij Represents the longitudinal force in the tire coordinate system. T ij This indicates the tire's output torque. R ij This indicates the tire size.

[0085] The lateral force in the tire coordinate system is determined by the tire model, and its calculation formula is as follows:

[0086]

[0087] in, For the front wheel steering angle, For the rear wheel steering angle, C f For the front wheel lateral stiffness, C r This refers to the rear wheel lateral stiffness.

[0088] Based on the current front and rear wheel steering angles of the vehicle, the longitudinal and lateral forces in the tire coordinate system can be converted to the vehicle body coordinate system. The calculation formula is as follows:

[0089]

[0090] in, For the front wheel steering angle, For the rear wheel steering angle, The transformation result representing the longitudinal force of each wheel from the tire coordinate system to the vehicle body coordinate system. The transformation result representing the lateral force of each wheel from the tire coordinate system to the vehicle coordinate system is ij∈{fl, fr, rl, rr}, which represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively.

[0091] Considering the spring-damper model of the suspension, the forces exerted by the suspension on the sprung mass include spring force, damping force, and active suspension force. Therefore, the formula for calculating the active suspension force is as follows:

[0092]

[0093] in, F act For active suspension force, F z Let N be the vertical force on the wheel, N be the spring constant, and c be the damping coefficient of the damper. z b The vertical displacement of the sprung mass can be calculated using sensors on the vehicle body. for z b The derivative; z s This refers to the vertical displacement of the suspension. for z s The derivative of .

[0094] Since vehicles with four-wheel independent drive and front-to-rear wheel steering are essentially overdrive systems, their driving degrees of freedom are greater than their vehicle attitude control degrees of freedom. Therefore, multiple force distribution methods exist. This application considers the optimal force distribution based on tire load rate, and the force distribution loss function... J The calculation is as follows:

[0095]

[0096] in, ω 1 and ω 2 are all weighting coefficients (i.e., other weights); μ i Let Var and E be the friction coefficient between each wheel and the road surface, respectively, representing the variance and mean of the tire load rate, i∈{fl, fr, rl, rr}; the force distribution loss function makes the load rates of each tire close, making full use of the adhesion of each wheel, and improving the vehicle stability margin with a lower overall vehicle load rate.

[0097] The loss function and other weights are allocated using the above force. ω 1 and ω 2. It is possible to calculate the minimum loss. F xi , F yi and F zi Then, by combining the force distribution control dynamics model, the hub motor torque, front and rear wheel steering angles, and active suspension forces are obtained.

[0098] S130, based on the expected control quantity of the vehicle and the current driving data of the vehicle, the initial weight value is optimized to obtain the optimized model predictive controller; wherein, the optimized model predictive controller is used to control the output control quantity of the vehicle during driving.

[0099] For example, in a specific embodiment of this application, since the initial weight values ​​are obtained by building a model, they may deviate from the expectations of real users. Therefore, an interactive feedback mechanism can be introduced to optimize the initial weight values ​​and obtain an optimized model prediction controller.

[0100] In some embodiments of this application, S130 may include: obtaining vehicle control index features corresponding to the current driving data of the vehicle; adjusting the initial weight value by calculating the error value between the expected control quantity of the vehicle and the vehicle control index features, as well as user behavior, to obtain the optimized model predictive controller.

[0101] For example, in a specific embodiment of this application, the current vehicle status is acquired in real time through the vehicle's built-in sensors (as a specific example of current vehicle driving data), and objective index characteristics (as a specific example of vehicle control index characteristics) are calculated. The error value between the objective index characteristics and the expected control quantity of the vehicle is calculated; simultaneously, the system continuously monitors the user's spontaneous intervention behavior (implicit user feedback), which is regarded as a feedback signal of dissatisfaction with the current control strategy. In addition, when encountering special road conditions (such as continuous bumps, sharp bends, etc.), the user can be slightly prompted by voice or HMI, and provided with "yes / no" or a short slider for user feedback (explicit user feedback) to obtain feedback signals. These feedback signals are the user behavior, which, combined with the error value, are used to fine-tune the weight matrix of the model predictive controller according to preset rules to obtain an optimized model predictive controller, thereby realizing the user's personalized driving needs.

[0102] In addition, in some embodiments of this application, the method for predictive control of vehicle lateral, longitudinal and vertical force coupling model may further include: acquiring user feedback behavior in response to in-vehicle prompts and monitoring the user's intervention behavior during driving; and correcting the subjective and objective quantitative model based on the feedback behavior and the intervention behavior.

[0103] Specifically, the system continuously monitors users' spontaneous intervention behaviors, considering these as interventions indicating dissatisfaction with the current control strategy. When encountering special road conditions (such as continuous bumps, sharp bends, etc.), the system provides subtle prompts to the user via voice or HMI, offering "yes / no" options or a brief slider for user feedback, thus obtaining feedback behavior. Based on the intervention and feedback behaviors, the subjective and objective quantitative model is revised to achieve adaptive optimization.

[0104] Simulation tests were conducted using the optimized model predictive controller obtained above, and the following results were obtained: Figure 2 and Figure 3 The simulation results are shown in the diagram. Figure 2 As can be seen from the embodiments described above, the actual trajectory and reference trajectory of the vehicle's lateral velocity can be almost identical, thereby achieving precise control of the vehicle's lateral velocity. By tracking the longitudinal velocity, the actual longitudinal velocity and the longitudinal reference velocity differ significantly during vehicle driving. This application allows for adjustment between the actual longitudinal velocity and the longitudinal reference velocity, causing the vehicle to approach the longitudinal reference velocity and maintain a stable driving state. Furthermore, by tracking the vehicle's yaw rate, the actual yaw rate can be almost identical to the reference value, achieving stable driving. Figure 3 Simulations of the roll and pitch control effects clearly show that the roll and pitch angles gradually move from an unstable state to a stable output state, meeting the driver's individual needs, demonstrating good control performance, and improving driving stability and reliability.

[0105] As can be seen from the above embodiments of this application, this application considers the coupling relationship of lateral, longitudinal and vertical dynamics, and achieves the global optimal performance of the vehicle under complex driving conditions through multi-degree-of-freedom collaborative optimization model predictive control, thereby improving the vehicle's handling stability, safety and ride comfort under extreme conditions; by constructing a quantitative mapping model between subjective evaluation and objective control indicators through a data-driven method, it overcomes the dependence of weight parameters on expert experience, enabling the control algorithm to adapt to different vehicle platforms, and significantly reducing calibration costs and development cycles; by introducing implicit and explicit user feedback mechanisms, the model predictive controller can learn online and adapt to the style preferences of different drivers, realizing a leap from "single performance" to "personalized driving experience for each driver".

[0106] Please refer to Figure 4 , Figure 4The diagram illustrates a block diagram of a vehicle lateral, longitudinal, and vertical force coupling model predictive control apparatus according to some embodiments of this application. It should be understood that this vehicle lateral, longitudinal, and vertical force coupling model predictive control apparatus corresponds to the method embodiments described above and is capable of performing the various steps involved in the method embodiments. The specific functions of this vehicle lateral, longitudinal, and vertical force coupling model predictive control apparatus can be found in the description above; detailed descriptions are omitted here to avoid repetition.

[0107] Figure 4 The vehicle lateral, longitudinal, and vertical force coupling model predictive control device includes at least one software functional module that can be stored in a memory or embedded in the vehicle lateral, longitudinal, and vertical force coupling model predictive control device in the form of software or firmware. The device includes: a weight acquisition module 410, used to acquire initial weight values ​​of the model predictive controller based on a subjective and objective quantitative model of vehicle driving and an optimization objective; wherein the subjective and objective quantitative model represents the mapping relationship between the driver's subjective evaluation data and the vehicle's motion state data; the model predictive controller includes a target control function and weights for different vehicle operating states; an expected control module 420, used to acquire the expected control quantity of the vehicle based on the initial weight values ​​and the model predictive controller; and an optimization control module 430, used to optimize the initial weight values ​​based on the expected control quantity of the vehicle and the vehicle's current driving data to obtain an optimized model predictive controller; wherein the optimized model predictive controller is used to control the output control quantity of the vehicle during driving.

[0108] Those skilled in the art will understand that, for the sake of convenience and brevity, the specific working process of the device described above can be referred to the corresponding process in the aforementioned method, and will not be elaborated further here.

[0109] Some embodiments of this application also provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, can perform the operation of any of the methods corresponding to the methods provided in the above embodiments.

[0110] Some embodiments of this application also provide a computer program product, which includes a computer program, wherein when the computer program is executed by a processor, it can implement the operation of any of the methods corresponding to the above embodiments provided in the above embodiments.

[0111] like Figure 5As shown, some embodiments of this application provide an electronic device 500, which includes a memory 510, a processor 520, and a computer program stored in the memory 510 and executable on the processor 520. When the processor 520 reads the program from the memory 510 via a bus 530 and executes the program, it can implement the methods of any of the above embodiments.

[0112] Processor 520 can process digital signals and can include various computing architectures. For example, it can be a complex instruction set computer architecture, a reduced instruction set computer architecture, or an architecture that implements multiple instruction set combinations. In some examples, processor 520 can be a microprocessor.

[0113] The memory 510 can be used to store instructions executed by the processor 520 or data related to the execution of instructions. These instructions and / or data may include code for implementing some or all of the functions of one or more modules described in the embodiments of this application. The processor 520 of this disclosure embodiment can be used to execute the instructions in the memory 510 to implement the methods shown above. The memory 510 includes dynamic random access memory, static random access memory, flash memory, optical memory, or other memories well known to those skilled in the art.

[0114] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application. It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures.

[0115] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0116] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

Claims

1. A method for predictive control of a vehicle using a coupled lateral, longitudinal, and vertical force model, characterized in that, include: Based on the subjective and objective quantitative model of vehicle driving and the optimization objective, the initial weight values ​​of the model predictive controller are obtained; wherein, the subjective and objective quantitative model represents the mapping relationship between the driver's subjective evaluation data and the vehicle motion state data; the model predictive controller includes the objective control function and the weights under different vehicle operating states; Based on the initial weight values ​​and the model predictive controller, the expected control quantity of the vehicle is obtained; Based on the expected control quantity of the vehicle and the current driving data of the vehicle, the initial weight value is optimized to obtain the optimized model predictive controller; wherein, the optimized model predictive controller is used to control the output control quantity of the vehicle during driving.

2. The method as described in claim 1, characterized in that, The method for obtaining initial weight values ​​for the model predictive controller based on the subjective and objective quantitative model of vehicle driving and the optimization objective includes: Using the preset score value corresponding to the vehicle model as the optimization objective, the subjective and objective quantification model is solved to obtain the initial weight value.

3. The method as described in claim 1 or 2, characterized in that, The step of obtaining the expected control quantity of the vehicle based on the initial weight value and the model predictive controller includes: Based on the throttle opening signal and steering wheel rotation signal of the vehicle, a desired trajectory with multiple parameters is predicted; wherein, the multiple parameters include the longitudinal velocity, lateral velocity, vertical velocity, yaw rate, pitch rate and roll rate of the vehicle. Based on the desired trajectory and a portion of the initial weight values, multiple control variables are obtained; The multiple control quantities are allocated using the weights other than the aforementioned partial weights in the initial weight values ​​and the model prediction controller to determine the expected control quantity of the vehicle.

4. The method as described in claim 3, characterized in that, The process involves obtaining multiple control variables based on the desired trajectory and a portion of the initial weight values, including: Construct the vehicle dynamics equations for the vehicle; A control loss function is constructed based on the vehicle dynamics equations; wherein, the control loss function includes the error loss between the desired trajectory and the current predicted trajectory, as well as the control quantity transformation loss; The optimal control quantity output by the control loss function is obtained by minimizing the loss and the partial weights; wherein the optimal control quantity is used as the plurality of control quantities.

5. The method as described in claim 3, characterized in that, The step of allocating the multiple control quantities using the weights other than the aforementioned partial weights in the initial weight values ​​and the model prediction controller to determine the expected control quantity of the vehicle includes: Construct a force distribution loss function related to tire load rate; The objective value of the force allocation loss function under minimum loss is obtained by using the other weights; The expected control quantity of the vehicle is obtained by using the target value and the force distribution control dynamics model in the model predictive controller.

6. The method as described in claim 1 or 2, characterized in that, The step of optimizing the initial weight values ​​based on the expected vehicle control quantity and the current vehicle driving data to obtain the optimized model predictive controller includes: Obtain the vehicle control index features corresponding to the current driving data of the vehicle. By calculating the error between the expected control quantity of the vehicle and the characteristics of the vehicle control index, as well as user behavior, the initial weight value is adjusted to obtain the optimized model predictive controller.

7. The method as described in claim 1 or 2, characterized in that, The subjective and objective quantitative model is constructed through the following steps: Obtain time series data of objective indicators of the vehicle under different operating conditions, and obtain subjective scores under the different operating conditions; The time series data is processed to obtain the feature values ​​of the objective indicators; The subjective-objective quantitative model is obtained based on the mapping relationship between the feature values ​​of all objective indicators and the subjective scores.

8. The method as described in claim 7, characterized in that, The method further includes: Acquire user feedback behavior in response to in-vehicle prompts and monitor the user's intervention behavior during driving; The subjective and objective quantitative model is corrected through the feedback behavior and the intervention behavior.

9. A device for predictive control of vehicle lateral, longitudinal, and vertical force coupling model, characterized in that, include: The weight acquisition module is used to acquire the initial weight values ​​of the model predictive controller based on the subjective and objective quantitative model of vehicle driving and the optimization objective; wherein, the subjective and objective quantitative model represents the mapping relationship between the driver's subjective evaluation data and the vehicle motion state data; the model predictive controller includes the target control function and the weights under different vehicle operating states; The expected control module is used to obtain the expected control quantity of the vehicle based on the initial weight value and the model prediction controller; An optimization control module is used to optimize the initial weight value based on the expected control quantity of the vehicle and the current driving data of the vehicle to obtain an optimized model predictive controller; wherein, the optimized model predictive controller is used to control the output control quantity of the vehicle during driving.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program, wherein the computer program, when executed by a processor, performs the method as described in any one of claims 1-8.

11. An electronic device, characterized in that, The method includes a memory, a processor, and a computer program stored in the memory and running on the processor, wherein the computer program is executed by the processor to perform the method as claimed in any one of claims 1-8.

Images