Stability control method for drive-by-wire chassis based on hybrid model for recognizing uncertainty

Abstract

This invention discloses a steerable chassis stability control method based on cognitive uncertainty in a hybrid model. The method includes: establishing a linear two-degree-of-freedom nominal vehicle dynamics model; constructing a residual model based on a Gaussian process, outputting the posterior mean and variance; integrating the two into a hybrid model, and mapping the posterior variance to a cognitive uncertainty coefficient; calculating the ideal center-of-gravity sideslip angle and yaw rate; dynamically correcting the stable region and adjusting the tracking weight of the ideal value in the phase plane by incorporating the cognitive uncertainty coefficient; designing an adaptive reaching law incorporating the cognitive uncertainty coefficient, and deriving the total sliding mode control law for calculating the additional yaw moment based on the hybrid model; and allocating the optimal driving torque to each wheel with the goal of minimizing tire load. This invention achieves dual adaptive adjustment of control gain and stability index by quantifying and utilizing the cognitive uncertainty of the model, effectively balancing strong robustness and control smoothness.

Description

Technical Field

[0001] This invention relates to the field of drive-by-wire chassis control technology, specifically to a drive-by-wire chassis stability control method based on a hybrid model for recognizing uncertainty. Background Technology

[0002] Chassis-by-wire yaw stability control is a key technology to ensure vehicle driving safety and handling stability. Its core objective is to suppress unstable trends such as oversteer and understeer in complex road conditions, emergency obstacle avoidance, low adhesion or extreme handling, so as to keep the yaw rate and center of gravity sideslip angle within a safe and controllable range, thereby maintaining good path following ability and lateral stability margin.

[0003] Among numerous control methods, sliding mode control is widely used in the field of drive-by-wire chassis control due to its relatively simple structure, ease of engineering implementation, and strong robustness against parameter perturbations and external disturbances. Traditional sliding mode controllers mostly rely on linear two-degree-of-freedom vehicle models for design. However, these models are usually based on assumptions such as small sideslip angles, linearized tire forces, and approximately constant longitudinal speeds, making it difficult to accurately describe the nonlinear dynamic characteristics of real vehicles under various operating conditions, which may lead to severe model mismatch.

[0004] To cover uncertainties caused by model mismatch, traditional sliding mode controllers often require a large fixed control gain. While a large gain setting can improve the approach speed, ensure fast convergence and strong disturbance rejection, it can sometimes lead to severe chattering and system overshoot. Conversely, if a smaller gain is chosen to improve control smoothness, it may be difficult to provide sufficient corrective torque and convergence rate when model mismatch is severe or disturbances are amplified, resulting in response hysteresis, reduced stability margin, and even the risk of vehicle instability in extreme cases. Therefore, fixed-gain sliding mode control struggles to balance robustness and control smoothness, necessitating the introduction of a control mechanism that can adaptively adjust the gain according to the level of uncertainty. Summary of the Invention

[0005] The purpose of this invention is to provide a method for controlling the stability of a drive-by-wire chassis based on a hybrid model to recognize uncertainty, in order to solve the problems that traditional sliding mode control cannot perceive uncertainty and uses fixed control gain, thus leading to the inability to balance robustness and control smoothness.

[0006] To achieve the above objectives, the present invention provides a method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, comprising the following steps:

[0007] S1: Establish a nominal vehicle dynamics model, which is a vehicle dynamics model based on the linear two-degree-of-freedom single-track assumption, and use a linear tire model to calculate the tire lateral force;

[0008] S2: Construct a residual model based on Gaussian process to learn the residual between the nominal vehicle dynamics model and the actual vehicle dynamics, and output the posterior predicted mean and posterior predicted variance of the residual between the state variables of the nominal vehicle dynamics model and the state variables of the actual vehicle.

[0009] S3: Integrate the nominal vehicle dynamics model and the residual model into a hybrid vehicle dynamics model, and map the posterior prediction variance output by the residual model to the cognitive uncertainty coefficient corresponding to the hybrid vehicle dynamics model;

[0010] S4: Based on driver input and vehicle driving status, calculate the vehicle's ideal center of gravity sideslip angle and ideal yaw rate using the nominal vehicle dynamics model;

[0011] S5: Based on the phase plane of the center of mass sideslip angle and the center of mass sideslip angular velocity, the stable region is divided according to the stability index. The stability index is corrected by the cognitive uncertainty coefficient and the tracking weight of the ideal center of mass sideslip angle and the ideal yaw angular velocity is dynamically adjusted.

[0012] S6: Define the total error tracking function and design the integral sliding surface according to the dynamically adjusted tracking weight; derive the integral sliding surface and select the adaptive approach law that incorporates the cognitive uncertainty coefficient; based on the hybrid vehicle dynamics model, the derivative of the integral sliding surface and the adaptive approach law, obtain the total control law, and calculate the expected additional yaw moment through the total control law;

[0013] S7: With minimizing the overall vehicle tire load rate as the optimization objective, the optimal driving torque for each wheel is calculated by taking the expected additional yaw moment, longitudinal force requirements, maximum motor torque, and road adhesion conditions as constraints, and then the optimal driving torque is distributed to each wheel hub motor of the drive-by-wire chassis.

[0014] To optimize the above technical solution, the specific measures also include:

[0015] Further, in step S1, the nominal vehicle dynamics model is:

[0016] ;

[0017] ;

[0018] in, It is the centroid sideslip angle; The steering angle of the front wheels; The vehicle's yaw rate; To add yaw moment; and These are the lateral stiffness of the front and rear wheels, respectively. For the overall vehicle weight; The longitudinal speed of the vehicle; Let be the moment of inertia about the center of mass; This is the distance from the front axle to the center of gravity. This is the distance from the rear axle to the center of mass; The centroid sideslip angle self-feedback matrix; This is the coupling matrix between the yaw rate and the sideslip angle of the center of mass; This is the coupling matrix between the sideslip angle of the center of mass and the yaw rate; This is the yaw rate self-feedback matrix; The input gain matrix is ​​the front wheel steering angle versus the center of gravity sideslip angle. Input the gain matrix for the front wheel steering angle versus yaw rate.

[0019] In step S2, the residual model based on the Gaussian process is constructed, and its expression is:

[0020] ;

[0021] in, For the centroid sideslip angle residual model; The posterior mean of the centroid sideslip angle residual; The posterior variance of the centroid sideslip angle residual; For the yaw rate residual model; The posterior mean of the yaw rate residual; The posterior variance of the yaw rate residual; is the input vector of the Gaussian process.

[0022] In step S3, the integration of the nominal vehicle dynamics model and the residual model into a hybrid vehicle dynamics model, and the mapping of the posterior prediction variance output by the residual model to the cognitive uncertainty coefficient corresponding to the hybrid vehicle dynamics model, specifically involves:

[0023] The expression for the hybrid vehicle dynamics model is:

[0024] ;

[0025] ;

[0026] Furthermore, the expression for the cognitive uncertainty coefficient is:

[0027] ;

[0028] in, The sideslip angular velocity of the center of mass in the hybrid vehicle dynamics model; For the yaw acceleration of the hybrid vehicle dynamics model; It represents the compensation value of the posterior mean of the centroid sideslip angle residual in continuous time form for the centroid sideslip angular velocity. This is the compensation value for the yaw acceleration based on the posterior mean of the yaw velocity residual in continuous time form. for The input vector at time step; For continuous time variables; Sampling time; The standard deviation of the signal of the Gaussian process kernel function for the centroid sideslip angle; denoted as the signal standard deviation of the kernel function for the yaw rate Gaussian process.

[0029] In step S4, the calculation of the vehicle's ideal center of gravity sideslip angle and ideal yaw rate using the nominal vehicle dynamics model specifically involves:

[0030] ;

[0031] in, The road surface adhesion coefficient; It is the acceleration due to gravity; This represents the stability coefficient.

[0032] Further, in step S5, the tracking weight expressions for the ideal centroid sideslip angle and the ideal yaw rate are respectively:

[0033] The tracking weight expression for the ideal centroid sideslip angle is:

[0034] ;

[0035] The tracking weight expression for the ideal yaw rate is: ;

[0036] in, This refers to the cognitive uncertainty coefficient. and The threshold for dividing the stable region, This is a stability indicator.

[0037] In step S6, the adaptive gain of the adaptive reaching law is determined in the following way:

[0038] Furthermore, the gain will be approached at a constant velocity. and exponentially approaching gain Replaced with an adaptive isokinetic reaching gain incorporating cognitive uncertainty coefficients and adaptive exponential approach gain :

[0039] ;

[0040] in, for Base gain; for The maximum gain reserve for cognitive uncertainty; for Gain coefficient; for Base gain; for The maximum gain reserve for cognitive uncertainty; for Gain coefficient.

[0041] Further, in step S6, based on the hybrid vehicle dynamics model, the derivative of the integral sliding surface, and the adaptive reaching law, the overall control law is obtained. The desired additional yaw moment is then calculated using the overall control law, specifically as follows:

[0042] ;

[0043] ; ;

[0044] in, To add a desired yaw moment; This is the overall control law; The ideal yaw acceleration; The ideal centroid sideslip angular velocity; The dimensionless normalization coefficient; For integral gain; For error tracking function; For integral sliding surfaces; Boundary layer thickness; It represents the compensation value of the posterior mean of the centroid sideslip angle residual in continuous time form for the centroid sideslip angular velocity. This is the compensation value for the yaw acceleration based on the posterior mean of the yaw velocity residual in continuous time form.

[0045] In step S7, the optimization objective is to minimize the overall vehicle tire load rate. This involves using the desired additional yaw moment, longitudinal force requirements, maximum motor torque, and road adhesion conditions as constraints to calculate the optimal driving torque for each wheel and distribute this optimal driving torque to each wheel hub motor of the drive-by-wire chassis. Specifically:

[0046] Vehicle tire load rate The expression is:

[0047] ;

[0048] Furthermore, the expected additional yaw moment, longitudinal force requirements, maximum motor torque, and road adhesion conditions are taken as constraints:

[0049] ;

[0050] Calculate the optimal driving torque for each wheel: ;

[0051] in, This provides the driving force for each wheel; The vertical force acting on each wheel; subscript These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. This represents the total longitudinal force of the vehicle. The wheel track; This is the maximum driving torque that the hub motor can produce; The radius is the wheel radius.

[0052] The present invention also proposes a computer-readable storage medium storing a computer program that enables a computer to execute the above-described method for controlling the stability of a drive-by-wire chassis based on hybrid model cognitive uncertainty.

[0053] Compared with the prior art, the beneficial effects of the present invention are:

[0054] This invention constructs a hybrid model that integrates physical mechanisms and data-driven approaches, and uses Gaussian process regression to perform nonlinear compensation on the residuals of the linear nominal model. This effectively captures unmodeled dynamics such as nonlinear changes in tire lateral stiffness, significantly improving the accuracy of vehicle dynamics models under complex operating conditions and providing a more reliable dynamic model foundation for controllers.

[0055] This invention innovatively quantifies the posterior variance of the Gaussian process output into a cognitive uncertainty coefficient, which is then used to correct the stability criterion based on the phase plane in real time. When the cognitive uncertainty of the model increases (such as when it is in an unknown operating condition), the stability constraint with the centroid sideslip angle as the core is automatically strengthened, and early intervention and conservative control are implemented. This overcomes the instability risks that may be caused by the lag in risk perception in traditional methods, and enhances the system's proactive safety capability in unfamiliar environments.

[0056] This invention corrects the phase plane stability index in real time by recognizing the uncertainty coefficient. When the model mismatch is severe and the uncertainty is large, the control gain is automatically increased to ensure fast convergence and strong anti-disturbance capability. When the model is accurate and the uncertainty is low, the gain is reduced to effectively suppress control chattering and avoid overshoot, thereby improving the control smoothness under normal operating conditions. This fundamentally solves the contradiction between robustness and control smoothness that traditional fixed-gain sliding mode control cannot balance.

[0057] The hybrid model structure, uncertainty mapping mechanism, and adaptive law design proposed in this invention provide a general approach that can be referenced for improving the performance of similar motion control systems, and have good application prospects and promotion value. Attached Figure Description

[0058] Figure 1 This is a schematic diagram of the process of the present invention.

[0059] Figure 2 This is a schematic diagram of the framework of the present invention. Detailed Implementation

[0060] The present invention will be further described in detail below through specific embodiments, but it should not be construed as limiting the scope of the subject matter of the present invention to the following embodiments. All technologies implemented based on the above content of the present invention fall within the scope of the present invention.

[0061] In some implementations, such as Figure 2 As shown, this invention provides a method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, comprising the following steps:

[0062] S1: Establish a nominal vehicle dynamics model, which is a vehicle dynamics model based on the linear two-degree-of-freedom single-track assumption, and use a linear tire model to calculate the tire lateral force.

[0063] Establish a vehicle dynamics model based on the linear two-degree-of-freedom single-track assumption:

[0064] ;

[0065] in, For the overall vehicle weight; This is the distance from the front axle to the center of gravity. This is the distance from the rear axle to the center of mass; The longitudinal speed of the vehicle; The vehicle's lateral speed; The vehicle's yaw rate; To add yaw moment; The lateral force is applied to the front axle tires. This refers to the lateral force exerted on the rear axle tires. Let be the moment of inertia about the center of mass.

[0066] The lateral force of the tire is calculated using a linear tire model:

[0067] ;

[0068] in, and These are the lateral stiffness of the front and rear wheels, respectively. It is the centroid sideslip angle; This refers to the steering angle of the front wheels.

[0069] Select the centroid side slip angle and vehicle yaw rate Let be the state variables, and the state-space equations are as follows:

[0070] ;

[0071] Among them, state variables Control quantity Input is .

[0072] In some implementations, the nominal vehicle dynamics model is represented as:

[0073] ;

[0074] ;

[0075] in, The centroid sideslip angle self-feedback matrix; This is the coupling matrix between the yaw rate and the sideslip angle of the center of mass; This is the coupling matrix between the sideslip angle of the center of mass and the yaw rate; This is the yaw rate self-feedback matrix; The input gain matrix is ​​the front wheel steering angle versus the center of gravity sideslip angle. Input the gain matrix for the front wheel steering angle versus yaw rate.

[0076] The nominal vehicle dynamics model ensures the interpretability of the controller's basic performance and physical mechanism under normal operating conditions.

[0077] S2: Construct a residual model based on a Gaussian process to learn the residual between the nominal vehicle dynamics model and the actual vehicle dynamics, and output the posterior predicted mean and posterior predicted variance of the residual between the state variables of the nominal vehicle dynamics model and the state variables of the actual vehicle.

[0078] In some implementations, data is collected based on CarSim and Simulink, and data is collected under different road surface adhesion coefficients and vehicle speeds, such as double lane change and single lane change.

[0079] Based on the physical characteristics of the vehicle's two-degree-of-freedom model, the following four-dimensional features are selected as the input vector for the Gaussian process. :

[0080] ;

[0081] The theoretical values ​​are calculated using the nominal vehicle dynamics model, and then subtracted from the collected actual values ​​to obtain the output target. and :

[0082] ;

[0083] in, for The residual of the centroid sideslip angle at any given moment; for The residual yaw rate at any given moment; for The true value of the centroid sideslip angle at any given moment; for Real value of yaw rate at any given moment; for The sideslip angle of the center of gravity calculated from the nominal vehicle dynamics model at any given moment; for The yaw rate is calculated from the nominal vehicle dynamics model at any given time.

[0084] In some implementations, the centroid sideslip angle residual and the yaw rate residual are trained separately, and the training set is as follows:

[0085] ;

[0086] in, For the set of input vectors; The set of output targets; For the above Moment ; For the above Moment or .

[0087] right and Z-score standardization is performed to prevent the kernel function from becoming overly sensitive to a certain dimension due to different units.

[0088] The posterior mean and posterior variance for each dimension are as follows:

[0089] ;

[0090] ;

[0091] in, This serves as the input during Gaussian process training; The posterior mean; For posterior variance; For kernel functions; Kernel vector; For the kernel matrix; To observe the noise variance; It is an identity matrix.

[0092] Choose the squared exponential kernel function (SE) as the kernel function for the Gaussian process:

[0093] ;

[0094] in, It is the characteristic length scaling matrix; parameters This represents the signal variance.

[0095] In some implementations, the model is trained using Matlab to solve for the optimal hyperparameter set. .

[0096] The trained residual model follows the distribution as follows:

[0097] ;

[0098] in, For the centroid sideslip angle residual model; The posterior mean of the centroid sideslip angle residual; The posterior variance of the centroid sideslip angle residual; For the yaw rate residual model; The posterior mean of the yaw rate residual; The posterior variance of the yaw rate residual; is the input vector of the Gaussian process.

[0099] S3: Integrate the nominal vehicle dynamics model and the residual model into a hybrid vehicle dynamics model, and map the posterior prediction variance output by the residual model to the cognitive uncertainty coefficient corresponding to the hybrid vehicle dynamics model.

[0100] The residual model is compensated on the continuous-time nominal vehicle dynamics model to obtain a continuous-time hybrid vehicle dynamics model:

[0101] ;

[0102] ;

[0103] Cognitive uncertainty coefficient The expression is:

[0104] ;

[0105] in, The angular velocity of the center of mass deflection; This is the yaw acceleration; It represents the compensation value of the posterior mean of the centroid sideslip angle residual in continuous time form for the centroid sideslip angular velocity. This is the compensation value for the yaw acceleration based on the posterior mean of the yaw velocity residual in continuous time form. for The input vector at time step; For continuous time variables; Sampling time; The standard deviation of the signal of the Gaussian process kernel function for the centroid sideslip angle; denoted as the signal standard deviation of the kernel function for the yaw rate Gaussian process.

[0106] S4: Based on driver input and vehicle driving status, calculate the vehicle's ideal center of gravity sideslip angle and ideal yaw rate using the nominal vehicle dynamics model.

[0107] In some implementations, the vehicle's sideslip angular velocity at the center of gravity during steady-state driving is... yaw acceleration ,make Substituting the state-space equations of the nominal vehicle dynamics model, we get:

[0108] ;

[0109] in, This is the stability coefficient.

[0110] The ideal sideslip angle is set to 0 to ensure vehicle stability at high speeds. Road surface adhesion conditions constrain the yaw rate. The ideal sideslip angle and ideal yaw rate ultimately used in controller design are expressed as:

[0111] ;

[0112] in, The road surface adhesion coefficient; This is the acceleration due to gravity.

[0113] S5: Based on the phase plane of the center of mass sideslip angle and the center of mass sideslip angular velocity, the stable region is divided according to the stability index. The stability index is corrected by the cognitive uncertainty coefficient, and the tracking weight of the ideal center of mass sideslip angle and the ideal yaw angular velocity is dynamically adjusted.

[0114] In some implementations, based on the plane of the sideslip angle and sideslip angular velocity, the stable region of the vehicle is divided into a stable region, a transition region, and an unstable region according to stability indices.

[0115] The expression for the stability index is:

[0116] ;

[0117] in, It is the slope coefficient of the stable boundary; when The area is considered stable, and vehicles are relatively safe. The tracking weight is zero; when As a transitional region, for The tracking weight gradually increases; when This is considered an unstable area, indicating the vehicle has become unstable. The tracking weight is 1; and The threshold for dividing the stable region.

[0118] definition To track the weights of the ideal centroid sideslip angle, :

[0119] ;

[0120] The phase plane stability index is modified by incorporating the cognitive uncertainty coefficient. As uncertainty increases, the phase plane stability index also increases, and the tracking weight for the ideal centroid sideslip angle also increases. Ultimately, the tracking weight for the ideal centroid sideslip angle... for:

[0121] ;

[0122] The tracking weight for the ideal yaw rate is: .

[0123] S6: Define the total error tracking function and design the integral sliding surface according to the dynamically adjusted tracking weight; derive the integral sliding surface and select the adaptive approach law that incorporates the cognitive uncertainty coefficient; based on the hybrid vehicle dynamics model, the derivative of the integral sliding surface and the adaptive approach law, obtain the total control law, and calculate the expected additional yaw moment through the total control law.

[0124] In some implementations, a total error tracking function is defined. :

[0125] ;

[0126] Design an integral sliding surface :

[0127] ;

[0128] in, The dimensionless normalization coefficient; For integral gain; The current moment; It is the integral variable.

[0129] For integral sliding surfaces Taking the derivative, we get :

[0130] ;

[0131] Selecting the exponential approach law:

[0132] ;

[0133] in, This is a constant-rate approach gain; It is a symbolic function; This is the exponentially approaching gain.

[0134] To reduce chattering, the aforementioned sign function is replaced with a saturation function. .

[0135] ;

[0136] in, This represents the boundary layer thickness.

[0137] In some implementations, the constant-rate approach gain is used. and exponentially approaching gain Replaced with an adaptive isokinetic reaching gain incorporating cognitive uncertainty coefficients and adaptive exponential approach gain :

[0138] ;

[0139] in, for Base gain; for The maximum gain reserve for cognitive uncertainty; for Gain coefficient; for Base gain; for The maximum gain reserve for cognitive uncertainty; for Gain coefficient.

[0140] The law of convergence is: .

[0141] In some implementations, let Assuming the vehicle state has reached the sliding surface, solve for the equivalent control law:

[0142] ;

[0143] in, The ideal yaw acceleration; The ideal centroid sideslip angular velocity;

[0144] make Solve the switching control law:

[0145] ;

[0146] The overall control law is The expected additional yaw moment is .

[0147] S7: With minimizing the overall vehicle tire load rate as the optimization objective, the optimal driving torque for each wheel is calculated by taking the expected additional yaw moment, longitudinal force requirements, maximum motor torque, and road adhesion conditions as constraints, and then the optimal driving torque is distributed to each wheel hub motor of the drive-by-wire chassis.

[0148] In some implementations, the desired additional yaw moment obtained by the sliding mode controller is used. The objective when distributing torque is to minimize the overall tire load rate. To optimize the objective:

[0149] ;

[0150] The constraints are: expected additional yaw moment, longitudinal force requirement, maximum motor torque, and road adhesion conditions.

[0151] ;

[0152] Calculate the optimal driving torque for each wheel: .

[0153] in, This provides the driving force for each wheel; The vertical force acting on each wheel; subscript These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. This represents the total longitudinal force of the vehicle. The wheel track; This is the maximum driving torque that the hub motor can produce; The radius is the wheel radius.

[0154] In another embodiment of the present invention, a computer-readable storage medium is provided storing a computer program that causes a computer to execute the above-described method for controlling the stability of a drive-by-wire chassis based on hybrid model cognitive uncertainty.

[0155] In the embodiments disclosed in this application, a computer storage medium may be a tangible medium that may contain or store programs for use by or in conjunction with an instruction execution system, apparatus, or device. The computer storage medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of computer storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fibers, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0156] In some embodiments, the flowchart of the present invention is as follows: Figure 1 As shown.

[0157] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Any simple modifications, equivalent substitutions, and improvements made by those skilled in the art to the above embodiments without departing from the scope of the technical solution of the present invention, based on the technical essence of the present invention, shall still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, characterized in that, Includes the following steps: S1: Establish a nominal vehicle dynamics model, which is a vehicle dynamics model based on the linear two-degree-of-freedom single-track assumption, and use a linear tire model to calculate the tire lateral force; S2: Construct a residual model based on Gaussian process to learn the residual between the nominal vehicle dynamics model and the actual vehicle dynamics, and output the posterior predicted mean and posterior predicted variance of the residual between the state variables of the nominal vehicle dynamics model and the state variables of the actual vehicle. S3: Integrate the nominal vehicle dynamics model and the residual model into a hybrid vehicle dynamics model, and map the posterior prediction variance output by the residual model to the cognitive uncertainty coefficient corresponding to the hybrid vehicle dynamics model; S4: Based on driver input and vehicle driving status, calculate the vehicle's ideal center of gravity sideslip angle and ideal yaw rate using the nominal vehicle dynamics model; S5: Based on the phase plane of the center of mass sideslip angle and the center of mass sideslip angular velocity, the stable region is divided according to the stability index. The stability index is corrected by the cognitive uncertainty coefficient and the tracking weight of the ideal center of mass sideslip angle and the ideal yaw angular velocity is dynamically adjusted. S6: Define the total error tracking function and design the integral sliding surface based on the dynamically adjusted tracking weights; The derivative of the integral sliding surface is obtained, and an adaptive reaching law incorporating cognitive uncertainty coefficients is selected. Based on the hybrid vehicle dynamics model, the derivative of the integral sliding surface, and the adaptive reaching law, the overall control law is obtained, and the expected additional yaw moment is calculated through the overall control law. S7: With minimizing the overall vehicle tire load rate as the optimization objective, the expected additional yaw moment, longitudinal force requirements, maximum motor torque, and road adhesion conditions are used as constraints to calculate the optimal driving torque for each wheel and distribute the optimal driving torque to each wheel hub motor of the drive-by-wire chassis.

2. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 1, is characterized in that: In step S1, the nominal vehicle dynamics model is: ； ； in, It is the centroid sideslip angle; The steering angle of the front wheels; The vehicle's yaw rate; To add yaw moment; and These are the lateral stiffness of the front and rear wheels, respectively. For the overall vehicle weight; The longitudinal speed of the vehicle; Let be the moment of inertia about the center of mass; This is the distance from the front axle to the center of gravity. This is the distance from the rear axle to the center of mass; The centroid sideslip angle self-feedback matrix; This is the coupling matrix between the yaw rate and the sideslip angle of the center of mass; This is the coupling matrix between the sideslip angle of the center of mass and the yaw rate; This is the yaw rate self-feedback matrix; The input gain matrix is ​​the front wheel steering angle versus the center of gravity sideslip angle. Input the gain matrix for the front wheel steering angle versus yaw rate.

3. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 1, is characterized in that: In step S2, the residual model based on the Gaussian process is constructed, and its expression is: ； in, For the centroid sideslip angle residual model; The posterior mean of the centroid sideslip angle residual; The posterior variance of the centroid sideslip angle residual; For the yaw rate residual model; The posterior mean of the yaw rate residual; The posterior variance of the yaw rate residual; is the input vector of the Gaussian process.

4. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 1, is characterized in that: In step S3, the integration of the nominal vehicle dynamics model and the residual model into a hybrid vehicle dynamics model, and the mapping of the posterior prediction variance output by the residual model to the cognitive uncertainty coefficient corresponding to the hybrid vehicle dynamics model, specifically involves: The expression for the hybrid vehicle dynamics model is: ； ； The expression for the cognitive uncertainty coefficient is: ； in, The sideslip angular velocity of the center of mass in the hybrid vehicle dynamics model; For the yaw acceleration of the hybrid vehicle dynamics model; It is the centroid sideslip angle; The steering angle of the front wheels; The vehicle's yaw rate; To add yaw moment; and These are the lateral stiffness of the front and rear wheels, respectively. Let be the moment of inertia about the center of mass; The centroid sideslip angle self-feedback matrix; This is the coupling matrix between the yaw rate and the sideslip angle of the center of mass; This is the coupling matrix between the sideslip angle of the center of mass and the yaw rate; This is the yaw rate self-feedback matrix; The input gain matrix is ​​the front wheel steering angle versus the center of gravity sideslip angle. Input the gain matrix for the front wheel steering angle versus yaw rate; It represents the compensation value of the posterior mean of the centroid sideslip angle residual in continuous time form for the centroid sideslip angular velocity. This is the compensation value for the yaw acceleration based on the posterior mean of the yaw velocity residual in continuous time form. for The input vector at time step; For continuous time variables; Sampling time; The standard deviation of the signal of the Gaussian process kernel function for the centroid sideslip angle; The standard deviation of the kernel function for the yaw rate Gaussian process; The posterior mean of the centroid sideslip angle residual; The posterior variance of the centroid sideslip angle residual; The posterior mean of the yaw rate residual; The posterior variance of the yaw rate residual is given.

5. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 1, is characterized in that: In step S4, the calculation of the vehicle's ideal sideslip angle and ideal yaw rate using the nominal vehicle dynamics model specifically involves: ； in, The ideal centroid sideslip angle; Ideal yaw rate; The road surface adhesion coefficient; It is the acceleration due to gravity; This is the distance from the front axle to the center of gravity. This is the distance from the rear axle to the center of mass; The longitudinal speed of the vehicle; Represents the stability coefficient; This refers to the steering angle of the front wheels.

6. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 1, is characterized in that: In step S5, the tracking weight expressions for the ideal centroid sideslip angle and the ideal yaw rate are respectively: The tracking weight expression for the ideal centroid sideslip angle is: ； The tracking weight expression for the ideal yaw rate is: ; in, This refers to the cognitive uncertainty coefficient. and The threshold for dividing the stable region, This is a stability indicator.

7. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 1, is characterized in that: In step S6, the adaptive gain of the adaptive reaching law is determined in the following way: The constant velocity approaches the gain and exponentially approaching gain Replaced with an adaptive isokinetic reaching gain incorporating cognitive uncertainty coefficients and adaptive exponential approach gain : ； in, for Base gain; for The maximum gain reserve for cognitive uncertainty; for Gain coefficient; for Base gain; for The maximum gain reserve for cognitive uncertainty; for Gain coefficient, This represents the cognitive uncertainty coefficient.

8. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 7, is characterized in that: In step S6, the overall control law is obtained based on the hybrid vehicle dynamics model, the derivative of the integral sliding surface, and the adaptive reaching law. The desired additional yaw moment is then calculated using the overall control law, specifically as follows: ； ； ； in, To add a desired yaw moment; This is the overall control law; Let be the moment of inertia about the center of mass; The centroid sideslip angle self-feedback matrix; This is the coupling matrix between the yaw rate and the sideslip angle of the center of mass; This is the coupling matrix between the sideslip angle of the center of mass and the yaw rate; This is the yaw rate self-feedback matrix; The input gain matrix is ​​the front wheel steering angle versus the center of gravity sideslip angle. Input the gain matrix for the front wheel steering angle versus yaw rate; It is the centroid sideslip angle; The steering angle of the front wheels; The vehicle's yaw rate; The tracking weight is the ideal centroid sideslip angle; The dimensionless normalization coefficient; For integral gain; For error tracking function; For integral sliding surfaces; Boundary layer thickness; It represents the compensation value of the posterior mean of the centroid sideslip angle residual in continuous time form for the centroid sideslip angular velocity. This is the compensation value for the yaw acceleration based on the posterior mean of the yaw velocity residual in continuous time form. The ideal yaw acceleration; The ideal centroid sideslip angular velocity.

9. The method for controlling the stability of a drive-by-wire chassis based on cognitive uncertainty using a hybrid model, as described in claim 1, is characterized in that: In step S7, the optimization objective is to minimize the overall vehicle tire load rate. This involves using the desired additional yaw moment, longitudinal force requirements, maximum motor torque, and road adhesion conditions as constraints to calculate the optimal driving torque for each wheel and distribute this optimal driving torque to each wheel hub motor of the drive-by-wire chassis. Specifically: Vehicle tire load rate The expression is: ； The constraints are: expected additional yaw moment, longitudinal force requirement, maximum motor torque, and road adhesion conditions. ； Calculate the optimal driving torque for each wheel: ; in, To add a desired yaw moment; This provides the driving force for each wheel; The vertical force acting on each wheel; The total longitudinal force of the vehicle; subscript These represent the left front wheel, right front wheel, left rear wheel, and right rear wheel, respectively. The wheel track; This is the maximum driving torque that the hub motor can produce; The radius of the wheel; This is the road surface adhesion coefficient.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The computer program causes the computer to execute the drive-by-wire chassis stability control method based on hybrid model cognitive uncertainty as described in any one of claims 1 to 9.

Images