Commercial vehicle driving control method considering load non-uniform distribution characteristic

Abstract

The invention discloses a commercial vehicle driving control method considering a load non-uniform distribution characteristic. The method comprises the following steps: acquiring and preprocessing sensor data of a commercial vehicle; estimating yaw disturbance torque caused by uneven load distribution in real time; a robust model prediction controller based on yaw disturbance torque compensation is constructed, and multi-axis coordinated yaw torque meeting the driving stability requirement under the non-uniform load working condition is output; driving force optimization distribution is carried out, and target driving force of each electric wheel is determined; and designing a driving force synchronous tracking controller based on the tire relaxation characteristic model, and outputting the torque of each electric wheel to complete the driving stability control of the commercial vehicle. According to the method, the yaw disturbance torque observer and robust model predictive control are fused, and the disturbance yaw torque is estimated in real time and compensated in a feedforward mode, so that pertinence and fineness are better in the aspect of disturbance suppression; the auxiliary sliding mode control can further improve the robustness of the system in an unknown or rapidly changing external environment.

Description

Technical Field

[0001] The present invention belongs to the technical field of automotive electronic control, and particularly relates to a driving control method for commercial vehicles considering the characteristics of uneven load distribution. Background Art

[0002] With the continuous improvement of the status of commercial vehicles in the transportation and logistics industries, the requirements for their reliability, economy, and efficiency are becoming increasingly stringent. As the core link of vehicle energy transmission and driving performance, the drive system is directly related to the power output and stability performance of commercial vehicles under various loads, road conditions, and working conditions. Based on this, distributed drive commercial vehicles are becoming a new trend in the industry's development. By configuring independent power units for each electric wheel or axle, the vehicle can flexibly distribute torque under complex working conditions of multiple wheels and axles, improve the power output efficiency and control accuracy, simplify the vehicle structure and optimize space utilization, and ultimately effectively reduce energy consumption while meeting the high-load requirements of commercial vehicles and improving the driving safety and economy of the vehicle.

[0003] However, commercial vehicles often suffer from uneven load distribution due to unbalanced loading or equipment defects, which in turn leads to yaw dynamic instability. In addition, due to the relaxation characteristics of tires, there is a varying degree of lag in the generation of longitudinal forces of electric wheels under uneven load distribution, further exacerbating the driving stability problem. When facing this typical non-ideal working condition of uneven load, existing technologies have tried to introduce methods such as sliding mode control, gain scheduling, and robust pole placement to improve control robustness and real-time performance, but most of them fail to simultaneously consider load disturbance estimation and multi-actuator synchronous coordination. Especially when facing rapid changes in parameters such as tire relaxation length and vertical load, the control accuracy and stability significantly decline.

[0004] Therefore, there is an urgent need for a driving control method for commercial vehicles that targets the characteristics of uneven load and takes into account the dynamic response of tire relaxation to improve the safety and reliability of commercial vehicles. Summary of the Invention

[0005] Aiming at the deficiencies of the above-mentioned existing technologies, the purpose of the present invention is to provide a driving control method for commercial vehicles considering the characteristics of uneven load distribution to solve the problem of decreased driving stability caused by ignoring the influence of uneven load characteristics and tire relaxation characteristics in the existing technologies.

[0006] To achieve the above purpose, the technical solution adopted by the present invention is as follows:

[0007] A driving control method for commercial vehicles considering the characteristics of uneven load distribution of the present invention comprises the following steps:

[0008] 1) Collect and preprocess the sensor data of the commercial vehicle, including the data of the centroid position, tire force, steering wheel angle, yaw angular velocity, and centroid sideslip angle;

[0009] 2) Design a yaw disturbing moment observer based on the multi-axle coupling dynamic model of commercial vehicles with uneven load distribution, and use the centroid position, tire forces, and yaw rate data collected in step 1) as the input of the observer to estimate in real time the yaw disturbing moment caused by uneven load distribution;

[0010] 3) According to the yaw disturbing moment obtained in step 2), construct a robust model predictive controller based on yaw disturbing moment compensation to achieve joint tracking control of the yaw rate and sideslip angle of the commercial vehicle, and output a multi-axle coordinated yaw moment that meets the driving stability requirements under uneven load conditions;

[0011] 4) Based on the multi-axle coordinated yaw moment obtained in step 3), combined with the adhesion characteristics of each tire of the commercial vehicle under different load distributions, optimize the distribution of the driving force to determine the target driving force of each electric wheel;

[0012] 5) Design a driving force synchronous tracking controller based on the tire relaxation characteristic model, output the torque of each electric wheel, and complete the driving stability control of the commercial vehicle.

[0013] Furthermore, the specific method for preprocessing the sensor data in step 1) is as follows:

[0014] Use the moving average method to perform smoothing filtering on the data including centroid position, tire forces, steering wheel angle, yaw rate, and sideslip angle of the centroid, eliminate the singular values in the data, and obtain a data set that meets the requirements.

[0015] Furthermore, the calculation formula of the moving average method is:

[0016]

[0017] where y t-n is the data before smoothing, y t is the data after smoothing, and n * is the number of filtering terms.

[0018] Furthermore, the specific design steps of the yaw disturbing moment observer in step 2) are as follows:

[0019] 21) Based on the multi-axle coupling dynamic model of commercial vehicles, extract the yaw disturbing moment caused by uneven load distribution, and perform frequency division processing on the yaw disturbing moment to extract the low-frequency and high-frequency components respectively;

[0020] 22) Design a yaw disturbing moment observer composed of a low-frequency disturbance observer and a high-frequency disturbance observer;

[0021] 23) Input the centroid position, tire force, and yaw angular velocity data collected in step 1) into the yaw disturbance torque observer to finally obtain the yaw disturbance torque.

[0022] Further, the frequency division processing of the yaw disturbance torque in step 21) is specifically as follows:

[0023] The commercial multi-axle coupling dynamics model under uneven load distribution is expressed as:

[0024]

[0025] where, Δx ij , Δy ij are respectively the longitudinal and lateral offsets of the centroid of the arm between the electric wheel force and the centroid, ΔF x,ij , ΔF y,ij are respectively the change amounts of the tire longitudinal force and the lateral force, M t is the yaw moment caused by considering the tire force asymmetry, M cg is the yaw moment caused by the centroid offset, d γ is the unobservable yaw moment disturbance, M dir is the driver input yaw moment, I z is the moment of inertia of the vehicle about the z-axis, γ is the yaw angular velocity, is the yaw angular acceleration;

[0026] Design a low-frequency observer and use a wide time window to estimate the yaw moment M cg caused by the centroid offset; design a high-frequency observer for observation and use the residual of the low-frequency observer as the input of the high-frequency observer to estimate the yaw moment M t caused by the tire force asymmetry.

[0027] Further, the yaw disturbance torque observer in step 22) is specifically designed as follows:

[0028] Design a low-frequency disturbance observer specifically as follows:

[0029]

[0030] where, t is the sampling time of the sensor; e1 is the estimated residual of the low-frequency disturbance observer, is the estimated value of the yaw angular velocity; β 11 , β 12 are the gain coefficients of the low-frequency disturbance observer, balancing the convergence speed and the anti-noise ability; is the estimated value of the yaw moment caused by the centroid offset; is the derivative with respect to the sensor sampling time; The estimated value of the vehicle's yaw rate that ignores the asymmetry of the tire force is the derivative with respect to the sensor sampling time; a x is the longitudinal acceleration of the vehicle; a y is the lateral acceleration of the vehicle; F z is the vertical force of the electric wheel; φ is a non - linear enhancement function that enhances the tracking ability of disturbance changes; α is a parameter of the non - linear enhancement function; δ is the maximum estimation error of e1; specifically, the non - linear enhancement function φ is designed as:

[0031]

[0032] where λ is a parameter of the non - linear enhancement function, and the parameter λ is designed to satisfy λ = k λ (1 - α), by adjusting the gain k λ to balance the fast response and smooth compensation;

[0033] Taking the estimated residual e1 in the low - frequency disturbance observer as the input, a high - frequency disturbance observer is designed. The specific design of the high - frequency disturbance observer is as follows:

[0034]

[0035] where ω ij represents the rotational speed of each electric wheel, represents the change rate of the rotational speed of each electric wheel; the high - frequency disturbance observer error is the estimated value of e1; is the estimated value of the vehicle's yaw rate that ignores the center - of - mass offset; is the estimated value of; β 21 , β 22 is the gain coefficient of the high - frequency disturbance observer; the non - linear function ψ is designed as:

[0036]

[0037] where δ2 is the maximum estimation error of e2; γ(e2) is the gain of the non - linear function, and γ0 is the nominal gain value.

[0038] Furthermore, the yaw disturbance torque finally obtained in step 23) is specifically expressed as follows:

[0039]

[0040] Furthermore, the specific steps of constructing a robust model predictive controller based on yaw disturbance torque compensation in step 3) are as follows:

[0041] 31) Based on the lateral dynamics model of a commercial vehicle under uneven load distribution conditions, establish a discretized state-space equation;

[0042] 32) Construct a two-degree-of-freedom reference model for the commercial vehicle to output the ideal yaw rate and the sideslip angle of the center of mass, which are used as the tracking reference targets of the controller;

[0043] 33) On the basis of considering the compensation of the yaw disturbing moment, design a nominal model predictive controller, and realize the prediction and control of the lateral state of the commercial vehicle by introducing a disturbance estimation compensation mechanism;

[0044] 34) Introduce an auxiliary sliding mode controller into the nominal model predictive controller, improve the anti-interference ability of the system through the design of robust control gain, and output a coordinated yaw moment.

[0045] Further, the discretized state-space equation established in step 31) is specifically as follows:

[0046] The lateral dynamics model of the commercial vehicle under uneven load distribution conditions is described as:

[0047]

[0048] where C1, C2, and C3 are the cornering stiffnesses of the front axle tires, middle axle tires, and rear axle tires respectively; a is the distance from the front axle to the center of gravity; b is the distance from the middle axle to the center of gravity; c is the distance between the middle axle and the rear axle; m is the vehicle mass; v is the vehicle speed; β is the sideslip angle of the vehicle center of mass, is the sideslip angular velocity of the vehicle center of mass;

[0049] Define the state variable χ = [β, γ] T , the input μ = M dir , the disturbance caused by the uneven load The system state equation is:

[0050]

[0051] where, is the derivative of the state variable with respect to the sensor sampling time; the system matrix the input matrix the disturbance matrix

[0052] Adopt a zero-order hold (ZOH), set the control input to be constant within the sampling period T, and discretize the system state equation to obtain the discretized state equation as follows:

[0053] χ(k + 1) = A d χ(k) + B d μ(k) + D d ω(k)

[0054] Among them, χ(k) is the state variable at the current moment k, and χ(k + 1) is the state variable at the next moment; μ(k) is the input at the current moment; ω(k) is the perturbation caused by the uneven load at the current moment; A d = e AT is the discretized system matrix, is the discretized input matrix, is the discretized perturbation matrix, τ is the temporary time variable used in the integration process, and is used to represent the time-varying system state during the integration process.

[0055] Furthermore, the specific steps of step 32) include:

[0056] The two-degree-of-freedom reference model of the commercial vehicle is specifically as follows:

[0057]

[0058] In the formula, v x is the longitudinal speed of the vehicle; δ f is the front wheel steering angle of the vehicle;

[0059] The ideal yaw angular velocity and the sideslip angle of the center of mass output are specifically as follows:

[0060] When the first derivative of the yaw rate is 0, the ideal yaw rate is controlled within the upper limit of the road adhesion coefficient, and the ideal yaw angular velocity γ d is expressed as:

[0061]

[0062] In the formula, μ is the road adhesion coefficient;

[0063] The ideal sideslip angle of the center of mass β d = 0.

[0064] Furthermore, the nominal model predictive controller in step 33) is specifically designed as follows:

[0065] Take the yaw disturbing moment as the input compensation of the controller, and the compensation moment is designed as:

[0066]

[0067] In the formula, α * is the design gain, which is used to adjust the compensation amount;

[0068] Through the compensation moment reduce the yaw disturbing moment The impact on vehicle dynamics and the restoration of the vehicle's motion to a relatively ideal state are defined. The nominal vehicle model is expressed as:

[0069]

[0070] Where ξ(k) is the nominal state of the system; ξ(k + 1) is the nominal state of the system at the next time point; is the input of the nominal model, u(k) is obtained by optimizing and solving the nominal model predictive controller;

[0071] Predict the future state based on the nominal model state variables of the vehicle at time k, and obtain:

[0072] ξ(k) = Πξ(k) + ΞU(k)

[0073] Where ξ(k) is the future system nominal state matrix, used to represent the set of nominal states of the vehicle at multiple future time points predicted starting from the current time k; the matrix Π is the state transition matrix, used to describe the state evolution law of the nominal vehicle model, indicating how the current state affects the future state; the matrix Ξ is the control gain matrix, used to describe the impact of the control quantity on the future state; the matrix U(k) is the control input matrix, representing the sequence of control inputs applied at multiple future time points starting from the current time k;

[0074] Adopt the soft constraint method and set the cost function for optimal control to effectively ensure the tracking of the ideal yaw rate and sideslip angle of the center of mass;

[0075] Define χ r (k) = [β d γ d T , and the cost function J(k) in the nominal model predictive control is expressed as:

[0076]

[0077] Where N p is the state prediction interval, N c is the control time domain; Q is the state weight matrix, R is the control input weight matrix, ε(k) is the relaxation constraint factor at time k, κ is the penalty coefficient, and i is the index variable of the predictive control step size; is the nominal state of the system at time k + i predicted at time k; is the nominal model input at time k + i predicted at time k;

[0078] The constraint range of the nominal model should be shrunk under the actual system constraint relationship and is expressed as:

[0079] ​

[0080] Among them, is the reduced model control input feasible region (constraint set); φ * is the gain matrix of the influence of the disturbance input on the nominal system; η(k) is the disturbance boundary of the nominal model; u max and u min are the upper limit value and the lower limit value of the control input respectively;

[0081] Expand and represent the objective function and combine the constraint conditions, and the form is as follows:

[0082]

[0083] In the formula,

[0084]

[0085] The optimal nominal control signal within the prediction range at the current moment is obtained by solving the quadratic programming problem in the optimal control sequence. The specific optimal control sequence is as follows:

[0086]

[0087] Take the first term * in the obtained optimal control sequence U as the nominal model control signal, and it is necessary to satisfy the constraint relationship

[0088] Furthermore, the auxiliary sliding mode controller in step 34) is specifically designed as follows:

[0089] Establish an error model to reflect the dynamic deviation between the actual system and the nominal model under the influence of external disturbances. The obtained state error equation is specifically:

[0090] ∈(k + 1) = A∈(k) + Bλ * η * (k) + ω * (k)

[0091] Among them, represents the error state between the actual system and the nominal model, and ∈(k + 1) is the error state at the k + 1 moment; λ * is the scale vector, η * (k) is the auxiliary sliding mode control law to ensure the convergence of the error state, and ω * (k) is the unobservable disturbance remaining after compensation ;

[0092] Design the discrete switching function s(k) according to the error ∈(k) between the actual system and the nominal model as:

[0093] s(k) = C e ∈(k)

[0094] where is the sliding mode surface coefficient;

[0095] The delay estimation method is used to measure the uncertainty disturbance ω * (k) in the state error equation, and the specific description is as follows:

[0096]

[0097] where is the estimated value of the disturbance at the current moment;

[0098] The discrete sliding mode reaching law is designed as:

[0099]

[0100] where α s , q s , t s , λ s , β s Here, the adjustment coefficients for defining the discrete sliding mode reaching law respectively satisfy 0 < α s < 1; 0 < q s t s < 1; 1 < λ s < β s ; σ(k) = ω(k) - ω(k - 1) represents the disturbance error within adjacent control periods;

[0101] Combined with the state error equation, the discrete switching function and the discrete sliding mode reaching law, the auxiliary sliding mode control law η * (k) is calculated as follows:

[0102]

[0103] Within the sampling interval t, the total control output of the driving stability controller is specifically:

[0104]

[0105] In the formula, is the output signal of the driving stability controller, that is, the coordinated yaw moment to ensure the driving stability of the commercial vehicle.

[0106] Furthermore, the specific steps of the driving force optimization distribution in step 4) are as follows:

[0107] 41) Construct the vehicle friction circle based on the road adhesion load of a single electric wheel, and use it as the boundary of the feasible region to design the driving force optimization function;

[0108] 42) Solve the driving force optimization function by satisfying the multi-axis coordinated yaw moment based on the constraints of the tire adhesion coefficient and the maximum output torque of the motor, and obtain the target driving force of each electric wheel.

[0109] Further, the driving force optimization function in step 41) is specifically designed as follows:

[0110] The road adhesion load η of a single electric wheel ij is represented by a friction ellipse:

[0111]

[0112] where F x,ij is the longitudinal force of each electric wheel, F y,ij is the lateral force of each electric wheel, F z,ij is the vertical force of each electric wheel, and μ ij is the tire adhesion coefficient of each electric wheel;

[0113] Considering the road adhesion load of a single electric wheel into the whole vehicle, the vehicle friction circle Φ is designed as follows:

[0114]

[0115] where n is the number of electric wheels;

[0116] The relationship between the longitudinal force F x,ij of each electric wheel and the lateral force F y,ij of each electric wheel is expressed as:

[0117]

[0118] where α ij is the tire slip angle of each electric wheel, C f,ij is the longitudinal stiffness of the tire, and C r,ij is the lateral stiffness of the tire;

[0119] Let Then the driving force optimization function J is specifically described as:

[0120]

[0121] Further, step 42) specifically includes:

[0122] The constraints of the tire adhesion coefficient and the maximum output torque of the motor are as follows:

[0123]

[0124] Among them, T max is the maximum torque that each electric wheel can output; R is the rolling radius of the electric wheel;

[0125] The solution process of the target driving force of each electric wheel is as follows:

[0126] While satisfying the constraints of the tire adhesion coefficient and the maximum output torque of the motor, it is necessary to satisfy the multi-axis coordinated yaw moment, specifically:

[0127]

[0128] Among them, F x,fl is the longitudinal force of the left front wheel, F x,fr is the longitudinal force of the right front wheel; F x,ml is the longitudinal force of the left middle wheel; F x,mr is the longitudinal force of the right middle wheel; F x,rl is the longitudinal force of the left rear wheel; F x,rr is the longitudinal force of the right rear wheel; B * is the wheelbase of the electric wheel;

[0129] Use a sequential quadratic programming solver to solve the following equations to obtain the target driving force of each electric wheel, as follows:

[0130]

[0131] Furthermore, the specific steps of designing the driving force synchronous tracking controller in step 5) are as follows:

[0132] 51) Establish a tire relaxation characteristic model to describe the hysteresis characteristic of the driving force response;

[0133] 52) Based on the target driving force obtained in step 4), design a dynamic-steady decoupled dual-reference model to generate a dynamic synchronization target value and a steady-state tracking target value;

[0134] 53) Introduce a composite error based on the dynamic synchronization and steady-state tracking target values in step 52), and solve it by designing a driving force synchronous tracking controller optimized based on a hybrid energy field to output the torque of the electric wheel that satisfies the driving force synchronous tracking control.

[0135] Furthermore, the specific steps of constructing the tire relaxation characteristic model in step 51) are as follows:

[0136] The tire relaxation longitudinal force equation is expressed as:

[0137]

[0138] In the formula, σ ijDenotes the relaxation length of each electric wheel, i.e., the distance the vehicle travels from when the tire starts to deform until it reaches a steady state; the superscript L indicates the inclusion of hysteresis effects; τ ij Denotes the time constant obtained by dividing the relaxation length of each electric wheel by the longitudinal speed of the vehicle;

[0139] Based on the tire relaxation longitudinal force equation, the tire relaxation characteristic model is specifically as follows:

[0140]

[0141] In the formula, I ω Is the moment of inertia of the electric wheel, T ij Is the driving torque of the in-wheel motor, T f,ij Is the rolling resistance torque, R e Is the effective rolling radius of the tire.

[0142] Furthermore, the designed dynamic-steady state decoupled dual reference model in step 52) is specifically as follows:

[0143] The dynamic synchronization target value generated based on the tire relaxation model Is:

[0144]

[0145] In the formula, Is the normalization factor used to map the dynamic response speeds of each electric wheel to a unified time scale, Is the geometric mean relaxation time; Is the target driving force obtained in step 4);

[0146] Under steady state conditions (i.e., when the longitudinal force change rate of each electric wheel approaches zero), the steady state tracking target value of each electric wheel is defined as As follows:

[0147]

[0148] In the formula, Is the reciprocal of the normalization factor.

[0149] Furthermore, step 53) is specifically as follows:

[0150] Based on the dynamic synchronization and steady state tracking target values in step 52), a composite error is introduced as follows:

[0151]

[0152] In the formula, the dynamic error term Is used to describe the deviation of the actual force F of the electric wheel x,ij Relative to the dynamic reference The steady state error term For describing the actual force F of the electric wheel x,ij and the steady-state reference deviation;

[0153] Based on the composite error, a hybrid energy function Ψ including synchronous potential energy, steady-state potential energy and dissipated energy is constructed and defined as follows:

[0154]

[0155] In the formula, K d , K s , K c are the dynamic stiffness coefficient, steady-state stiffness coefficient and high-frequency suppression coefficient respectively;

[0156] Based on the hybrid energy function Ψ, the Lyapunov function is defined as:

[0157]

[0158] Taking the derivative of V ij gives:

[0159]

[0160] To ensure The torque of the electric wheel that satisfies the driving force synchronous tracking control is solved as follows:

[0161]

[0162] In the formula, K v >0 is the error adjustment gain, represents the gradient direction of the hybrid energy field function in the error space.

[0163] Advantages of the present invention:

[0164] In the present invention, a yaw disturbance torque observer and a robust model predictive control are integrated. By estimating and feedforward compensating the disturbance yaw torque in real time, it is more targeted and precise in disturbance suppression; at the same time, the auxiliary sliding mode control can further improve the system robustness in an unknown or rapidly changing external environment, overcoming the deficiencies of strong model dependence in traditional methods and easy occurrence of compensation lag when the disturbance intensity is large.

[0165] The driving force synchronous tracking control of the present invention adopts a method combining dynamic-steady-state decoupling and energy field optimization, which can not only guide the phase synchronization of each electric wheel in the transient stage, reduce the force output hysteresis caused by the difference in tire relaxation characteristics, but also gradually approach the target torque of the optimal distribution in the steady state, realizing precise control and efficient utilization of the output force of each electric wheel; compared with the traditional single-wheel independent adjustment or fixed ratio scheme, the method of the present invention can significantly improve the cooperation efficiency and control accuracy among multiple wheels. Description of the Drawings

[0166] Figure 1 This is a flowchart of the method of the present invention.

[0167] Figure 2 This is a block diagram of the driving force synchronous tracking control of the present invention. Detailed Embodiments

[0168] For the convenience of those skilled in the art to understand, the present invention will be further described below in conjunction with the embodiments and the drawings. The content mentioned in the embodiments does not limit the present invention.

[0169] Refer to Figure 1 、 Figure 2 As shown, a commercial vehicle drive control method considering the characteristics of uneven load distribution of the present invention is as follows:

[0170] 1) Collect and preprocess the sensor data of the commercial vehicle, including the data of the centroid position, tire force, steering wheel angle, yaw rate and centroid side slip angle;

[0171] Specifically, the specific method for preprocessing the sensor data in step 1) is as follows:

[0172] Use the moving average method to perform smoothing filtering on the data of the centroid position, tire force, steering wheel angle, yaw rate and centroid side slip angle, eliminate the singular values in the data and obtain a data set that meets the requirements;

[0173] The calculation formula of the moving average method is:[[]]

[0174]

[0175] where y t-n is the data before smoothing, y t is the data after smoothing, and n * is the number of filtering terms.

[0176] 2) Design a yaw disturbing moment observer based on the multi-axle coupling dynamic model of the commercial vehicle with uneven load distribution, and use the centroid position, tire force and yaw rate data collected in step 1) as the input of the observer to estimate the yaw disturbing moment caused by uneven load distribution in real time;

[0177] Specifically, the specific design steps of the yaw disturbing moment observer in step 2) are as follows:[[]]

[0178] 21) Based on the multi-axle coupling dynamic model of the commercial vehicle, extract the yaw disturbing moment caused by uneven load distribution, and perform frequency division processing on the yaw disturbing moment to extract the low-frequency and high-frequency components respectively;

[0179] 22) Design a yaw disturbance torque observer composed of a low-frequency disturbance observer and a high-frequency disturbance observer;

[0180] 23) Input the centroid position, tire force, and yaw angular velocity data collected in step 1) into the yaw disturbance torque observer to finally obtain the yaw disturbance torque.

[0181] Among them, the frequency division processing of the yaw disturbance torque in step 21) is specifically as follows:

[0182] The commercial multi-axis coupling dynamics model under uneven load distribution is expressed as:

[0183]

[0184] Among them, Δx ij , Δy ij are respectively the longitudinal and lateral offsets of the centroid of the force arm between the electric wheel force and the centroid, ΔF x,ij , ΔF y,ij are respectively the change amounts of the tire longitudinal force and the lateral force, M t is the yaw torque caused by considering the tire force asymmetry, M cg is the yaw torque caused by the centroid offset, d γ is the unobservable yaw torque disturbance, M dir is the driver input yaw torque, I z is the moment of inertia of the vehicle about the z-axis, γ is the yaw angular velocity, is the yaw angular acceleration;

[0185] Design a low-frequency observer and use a wide time window to estimate the yaw torque M cg caused by the centroid offset; Design a high-frequency observer for observation and use the residual of the low-frequency observer as the input of the high-frequency observer to estimate the yaw torque M t caused by the tire force asymmetry.

[0186] Among them, the yaw disturbance torque observer in step 22) is specifically designed as follows:

[0187] The specific design of the low-frequency disturbance observer is as follows:

[0188]

[0189] Among them, t is the sampling time of the sensor; e1 is the estimated residual of the low-frequency disturbance observer, is the yaw angular velocity estimated value; β 11 , β 12 are the gain coefficients of the low-frequency disturbance observer, balancing the convergence speed and the anti-noise ability; is the estimated value of the yaw moment caused by the centroid offset; is the derivative with respect to the sensor sampling time; is the estimated value of the vehicle yaw rate neglecting the tire force asymmetry, is the derivative with respect to the sensor sampling time; a x is the longitudinal acceleration of the vehicle; a y is the lateral acceleration of the vehicle; F z is the vertical force of the electric wheel; φ is the non - linear enhancement function to enhance the tracking ability of disturbance changes; α is the parameter of the non - linear enhancement function; δ is the maximum estimation error of e1; specifically, the non - linear enhancement function φ is designed as:

[0190]

[0191] where λ is the parameter of the non - linear enhancement function, and the parameter λ is designed to satisfy λ = k λ (1 - α), by adjusting the gain k λ to balance the fast response and smooth compensation;

[0192] Design a high - frequency disturbance observer with the estimated residual e1 in the low - frequency disturbance observer as the input. The specific design of the high - frequency disturbance observer is as follows:

[0193]

[0194] where ω ij represents the rotational speed of each electric wheel, represents the change rate of the rotational speed of each electric wheel; the high - frequency disturbance observer error is the estimated value of e1; is the estimated value of the vehicle yaw rate neglecting the centroid offset; is the estimated value of; β 21 , β 22 is the gain coefficient of the high - frequency disturbance observer; the non - linear function ψ is designed as:

[0195]

[0196] where δ2 is the maximum estimation error of e2; γ(e2) is the gain of the non - linear function, and γ0 is the nominal gain value.

[0197] where the final yaw disturbance moment obtained in step 23) is specifically expressed as follows:

[0198]

[0199] 3) Based on the yaw disturbance moment obtained in step 2), construct a robust model predictive controller based on yaw disturbance moment compensation to achieve joint tracking control of the yaw rate and sideslip angle of the center of mass of the commercial vehicle, and output a multi-axis coordinated yaw moment that meets the driving stability requirements under uneven load conditions;

[0200] Specifically, the specific steps for constructing the robust model predictive controller based on yaw disturbance moment compensation in step 3) are as follows:

[0201] 31) Based on the lateral dynamic model of the commercial vehicle under uneven load distribution, establish a discretized state-space equation;

[0202] 32) Construct a two-degree-of-freedom reference model for the commercial vehicle to output the ideal yaw rate and sideslip angle of the center of mass as the tracking reference target of the controller;

[0203] 33) On the basis of considering yaw disturbance moment compensation, design a nominal model predictive controller, and through the introduction of a disturbance estimation compensation mechanism, achieve the prediction and control of the lateral state of the commercial vehicle;

[0204] 34) Introduce an auxiliary sliding mode controller into the nominal model predictive controller, improve the anti-interference ability of the system through robust control gain design, and output a coordinated yaw moment.

[0205] Among them, the discretized state-space equation established in step 31) is specifically as follows:

[0206] The lateral dynamic model of the commercial vehicle under uneven load distribution is described as:

[0207]

[0208] Among them, C1, C2, and C3 are the cornering stiffnesses of the front axle tires, middle axle tires, and rear axle tires respectively; a is the distance from the front axle to the center of gravity; b is the distance from the middle axle to the center of gravity; c is the distance between the middle axle and the rear axle; m is the vehicle mass; v is the vehicle speed; β is the sideslip angle of the vehicle center of mass, is the sideslip angular velocity of the vehicle center of mass;

[0209] Define the state variable χ = [β, γ] T , the input μ = M dir , the disturbance caused by uneven load The system state equation is:

[0210]

[0211] Among them, is the derivative of the state variable with respect to the sensor sampling time; the system matrix input matrix Disturbance matrix

[0212] Using a zero-order hold (ZOH), setting the control input to be constant within the sampling period T, discretizing the system state equation, and obtaining the discretized state equation as follows:

[0213] χ(k + 1) = A d χ(k) + B d μ(k) + D d ω(k)

[0214] where χ(k) is the state variable at the current time k, χ(k + 1) is the state variable at the next time; μ(k) is the input at the current time; ω(k) is the disturbance caused by the uneven load at the current time; A d = e AT is the discretized system matrix, is the discretized input matrix, is the discretized disturbance matrix, τ is the temporary time variable used in the integration process to represent the system state varying with time.

[0215] Among them, the specific steps of step 32) include:

[0216] The two-degree-of-freedom reference model of the commercial vehicle is specifically as follows:

[0217]

[0218] In the formula, v x is the longitudinal speed of the vehicle; δ f is the front wheel steering angle of the vehicle;

[0219] The ideal yaw rate and the sideslip angle of the center of mass output are specifically as follows:

[0220] When the first derivative of the yaw rate is 0, the ideal yaw rate is controlled within the upper limit of the road adhesion coefficient, and the ideal yaw angular velocity γ d is expressed as:

[0221]

[0222] In the formula, μ is the road adhesion coefficient;

[0223] The ideal sideslip angle of the center of mass β d = 0.

[0224] Among them, the nominal model predictive controller in step 33) is specifically designed as follows:

[0225] Taking the yaw disturbing moment as the input compensation of the controller, the compensation moment is designed as:

[0226]

[0227] where α * is the designed gain for adjusting the compensation amount;

[0228] By the compensation torque reduce the yaw disturbance torque caused by uneven load on the vehicle dynamics, and make the vehicle motion return to a relatively ideal state. Define the vehicle nominal model as:

[0229]

[0230] where ξ(k) is the nominal state of the system; ξ(k + 1) is the nominal state of the system at the next time point; is the nominal model input, u(k) is obtained by optimizing and solving with the nominal model predictive controller;

[0231] Predict the future state according to the nominal model state variables of the vehicle at time k, and get:

[0232] ξ(k) = Πξ(k) + ΞU(k)

[0233] where ξ(k) is the future system nominal state matrix, used to represent the set of vehicle nominal states at multiple future time points predicted starting from the current time k; the matrix Π is the state transition matrix, used to describe the state evolution law of the vehicle nominal model, indicating how the current state affects the future state; the matrix Ξ is the control gain matrix, used to describe the influence of the control quantity on the future state; the matrix U(k) is the control input matrix, representing the sequence of control inputs applied at multiple future time points starting from the current time k;

[0234] Adopt the soft constraint method and set the cost function for optimal control to effectively ensure the tracking of the ideal yaw rate and sideslip angle of the center of mass;

[0235] Define χ r (k) = [β d γ d T , and the cost function J(k) in the nominal model predictive control is expressed as:

[0236]

[0237] where N p is the state prediction interval, N c ​Let \(t\) be the control time domain; \(Q\) be the state weight matrix, \(R\) be the control input weight matrix, \(\varepsilon(k)\) be the slack constraint factor at time \(k\), \(\kappa\) be the penalty coefficient, and \(i\) be the index variable of the prediction control step size; is the nominal state of the system at time \(k + i\) predicted at time \(k\); is the nominal model input at time \(k + i\) predicted at time \(k\);

[0238] The nominal model constraint range should be shrunk under the actual system constraint relationship as follows:

[0239]

[0240] where is the shrunk feasible region (constraint set) of the model control input; \(\varphi\) * is the gain matrix of the influence of the disturbance input on the nominal system; \(\eta(k)\) is the disturbance boundary of the nominal model; \(u\) max and \(u\) min are the upper limit value and lower limit value of the control input respectively;

[0241] The objective function is expanded and expressed in combination with the constraint conditions in the following form:

[0242]

[0243] In the formula,

[0244]

[0245] The optimal nominal control signal within the prediction range at the current moment is obtained by solving the quadratic programming problem in the optimal control sequence. The optimal control sequence is specifically as follows:

[0246]

[0247] The first term * in the obtained optimal control sequence \(U\) (k) is used as the nominal model control signal and needs to satisfy the constraint relationship

[0248] where the auxiliary sliding mode controller in step 34) is specifically designed as follows:

[0249] An error model is established to reflect the dynamic deviation between the actual system and the nominal model under the influence of external disturbances. The obtained state error equation is specifically:

[0250] \(\in(k + 1)=A\in(k)+B\lambda\) * \(\eta\) * (k)+\(\omega\) * (k)

[0251] Among them, represents the error state between the actual system and the nominal model, and ∈(k + 1) is the error state at time k + 1; λ * is the scale vector, and η * (k) is the auxiliary sliding mode control law to ensure the convergence of the error state, and ω * (k) is the unobservable disturbance remaining after compensation is completed;

[0252] According to the error ∈(k) between the actual system and the nominal model, the discrete switching function s(k) is designed as:

[0253] s(k) = C e ∈(k)

[0254] Among them, is the sliding mode surface coefficient;

[0255] The delayed estimation method is adopted to measure the uncertainty disturbance ω * (k) in the state error equation, and the specific description is as follows:

[0256]

[0257] Among them, is the estimated value of the disturbance at the current moment;

[0258] The discrete sliding mode reaching law is designed as:

[0259]

[0260] Among them, α s , q s , t s , λ s , β s Here are the adjustment coefficients for defining the discrete sliding mode reaching law, which respectively satisfy 0 < α s < 1; 0 < q s t s < 1; 1 < λ s < β s ; σ(k) = ω(k) - ω(k - 1) represents the disturbance error within adjacent control periods;

[0261] Combined with the state error equation, the discrete switching function and the discrete sliding mode reaching law, the auxiliary sliding mode control law η * (k) is obtained as follows:

[0262]

[0263] Within the sampling interval t, the total control output of the drive stability controller is specifically:

[0264]

[0265] In the formula, is the output signal of the drive stability controller, that is, the coordinated yaw moment to ensure the drive stability of the commercial vehicle.

[0266] 4) Based on the multi-axle coordinated yaw moment obtained in step 3), combined with the adhesion characteristics of each tire of the commercial vehicle under different load distributions, optimize the distribution of the driving force to determine the target driving force of each electric wheel;

[0267] Specifically, the specific steps of the driving force optimization distribution in step 4) are as follows:

[0268] 41) Construct a vehicle friction circle based on the road adhesion load of a single electric wheel, and use it as the boundary of the feasible region to design a driving force optimization function;

[0269] 42) Solve the driving force optimization function while satisfying the multi-axle coordinated yaw moment based on the constraints of the tire adhesion coefficient and the maximum output torque of the motor to obtain the target driving force of each electric wheel.

[0270] Among them, the driving force optimization function in step 41) is specifically designed as follows:

[0271] The road adhesion load η of a single electric wheel ij is represented by a friction ellipse:

[0272]

[0273] Among them, F x,ij is the longitudinal force of each electric wheel, F y,ij is the lateral force of each electric wheel, F z,ij is the vertical force of each electric wheel, and μ ij is the tire adhesion coefficient of each electric wheel;

[0274] Considering the road adhesion load of a single electric wheel into the whole vehicle, the vehicle friction circle Φ is designed as follows:

[0275]

[0276] Among them, n is the number of electric wheels;

[0277] The relationship between the longitudinal force F x,ij of each electric wheel and the lateral force F y,ij of each electric wheel is expressed as:

[0278]

[0279] Among them, α ijis the tire slip angle of each electric wheel, C f,ij is the longitudinal stiffness of the tire, C r,ij is the lateral stiffness of the tire;

[0280] Let Then the driving force optimization function J is specifically described as:

[0281]

[0282] Among them, the specific steps of step 42) include:

[0283] The constraints of the tire adhesion coefficient and the maximum output torque of the motor are specifically as follows:

[0284]

[0285] Among them, T max is the maximum torque that each electric wheel can output; R is the rolling radius of the electric wheel;

[0286] The solving process of the target driving force of each electric wheel is specifically as follows:

[0287] While satisfying the constraints of the tire adhesion coefficient and the maximum output torque of the motor, it is necessary to satisfy the multi-axis coordinated yaw moment, specifically:

[0288]

[0289] Among them, F x,fl is the longitudinal force of the left front wheel, F x,fr is the longitudinal force of the right front wheel; F x,ml is the longitudinal force of the left middle wheel; F x,mr is the longitudinal force of the right middle wheel; F x,rl is the longitudinal force of the left rear wheel; F x,rr is the longitudinal force of the right rear wheel; B * is the wheelbase of the electric wheel;

[0290] Use the sequential quadratic programming solver to solve the following equation to obtain the target driving force of each electric wheel, as follows:

[0291]

[0292] 5) Design a driving force synchronous tracking controller based on the tire relaxation characteristic model, output the torque of each electric wheel, and complete the driving stability control of the commercial vehicle;

[0293] Specifically, the specific steps of designing the driving force synchronous tracking controller in step 5) are as follows:

[0294] 51) Establish a tire relaxation characteristic model to describe the hysteresis characteristic of the driving force response;

[0295] 52) Based on the target driving force obtained in step 4), design a dynamic-steady decoupled dual-reference model to generate a dynamic synchronization target value and a steady-state tracking target value;

[0296] 53) Introduce a composite error based on the dynamic synchronization and steady-state tracking target values in step 52), and solve it by designing a driving force synchronization tracking controller optimized based on a hybrid energy field to output the motor wheel torque that satisfies the driving force synchronization tracking control.

[0297] Among them, the construction of the tire relaxation characteristic model in step 51) is specifically as follows:

[0298] The tire relaxation longitudinal force equation is expressed as:

[0299]

[0300] In the formula, σ ij represents the relaxation length of each motor wheel, that is, the distance traveled by the vehicle from the start of tire deformation to reaching the steady state; the superscript L represents the inclusion of the hysteresis effect; τ ij represents the time constant obtained by dividing the relaxation length of each motor wheel by the longitudinal speed of the vehicle;

[0301] Based on the tire relaxation longitudinal force equation, the tire relaxation characteristic model is specifically as follows:

[0302]

[0303] In the formula, I ω is the moment of inertia of the motor wheel, T ij is the driving torque of the hub motor, T f,ij is the rolling resistance torque, R e is the effective rolling radius of the tire.

[0304] Among them, the designed dynamic-steady decoupled dual-reference model in step 52) is specifically as follows:

[0305] The dynamic synchronization target value generated based on the tire relaxation model is:

[0306]

[0307] In the formula, is the normalization factor used to map the dynamic response speeds of each motor wheel to a unified time scale, is the geometric mean relaxation time; is the target driving force obtained in step 4);

[0308] Under steady-state conditions (that is, the longitudinal force change rate of each motor wheel approaches zero), define the steady-state tracking target value of each motor wheel as as follows:

[0309]

[0310] In the formula, is the reciprocal of the normalization factor.

[0311] Among them, the specific steps of step 53) are as follows:

[0312] Based on the dynamic synchronization and steady-state tracking target values in step 52), a composite error is introduced as follows:

[0313]

[0314] In the formula, the dynamic error term is used to describe the deviation of the actual force F of the electric wheel x,ij relative to the dynamic reference ; the steady-state error term is used to describe the deviation of the actual force F of the electric wheel x,ij from the steady-state reference .

[0315] Based on the composite error, a hybrid energy function Ψ including synchronous potential energy, steady-state potential energy and dissipated energy is constructed and defined as follows:

[0316]

[0317] In the formula, K d , K s , K c are the dynamic stiffness coefficient, steady-state stiffness coefficient and high-frequency suppression coefficient respectively;

[0318] Based on the hybrid energy function Ψ, the Lyapunov function is defined as:

[0319]

[0320] Taking the derivative of V ij yields:

[0321]

[0322] To ensure The torque of the electric wheel that satisfies the driving force synchronous tracking control is solved as follows:

[0323]

[0324] In the formula, K v > 0 is the error adjustment gain, represents the gradient direction of the hybrid energy field function in the error space.

[0325] The specific application scenarios of the present invention are numerous. The above description is only the preferred implementation mode of the present invention. It should be noted that for those of ordinary skill in the art, several improvements can be made without departing from the principle of the present invention, and these improvements should also be regarded as the protection scope of the present invention.

Claims

1. A drive control method for commercial vehicles considering the characteristics of uneven load distribution, characterized in that, The steps are as follows: 1) Collect and preprocess the sensor data of the commercial vehicle, including the data of the centroid position, tire force, steering wheel angle, yaw angular velocity, and centroid sideslip angle; 2) Design a yaw disturbance torque observer based on the multi-axle coupling dynamic model of the commercial vehicle with uneven load distribution, and use the centroid position, tire force, and yaw angular velocity data collected in step 1) as the input of the observer to estimate in real time the yaw disturbance torque caused by uneven load distribution; 3) According to the yaw disturbance torque obtained in step 2), construct a robust model predictive controller based on yaw disturbance torque compensation to realize the joint tracking control of the yaw angular velocity and centroid sideslip angle of the commercial vehicle, and output the multi-axle coordinated yaw torque that meets the driving stability requirements under the uneven load condition; 4) Based on the multi-axle coordinated yaw torque obtained in step 3), combined with the adhesion characteristics of each tire of the commercial vehicle under different load distributions, optimize the distribution of the driving force, and determine the target driving force of each electric wheel; 5) Design a driving force synchronous tracking controller based on the tire relaxation characteristic model, output the torque of each electric wheel, and complete the driving stability control of the commercial vehicle.

2. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 1, characterized in that The specific method for preprocessing the sensor data in step 1) is as follows: Use the moving average method to perform smoothing filtering on the data including the centroid position, tire force, steering wheel angle, yaw angular velocity, and centroid sideslip angle data, eliminate the singular values in the data, and obtain a data set that meets the requirements; The calculation formula of the moving average method is: where y t-n is the data before smoothing, and y t is the data after smoothing, and n * is the number of filtering terms.

3. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 1, characterized in that The specific design steps of the yaw disturbance torque observer in step 2) are as follows: 21) Based on the multi-axle coupling dynamic model of the commercial vehicle, extract the yaw disturbance torque caused by uneven load distribution, and perform frequency division processing on the yaw disturbance torque to extract the low-frequency and high-frequency components respectively; 22) Design a yaw disturbance torque observer composed of a low-frequency disturbance observer and a high-frequency disturbance observer; 23) Input the centroid position, tire force, and yaw angular velocity data collected in step 1) into the yaw disturbance torque observer, and finally obtain the yaw disturbance torque.

4. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 3, characterized in that The specific frequency division processing of the yaw disturbance torque in step 21) is as follows: The multi-axle coupling dynamic model of the commercial vehicle under uneven load distribution is expressed as: Among them, Δx ij , Δy ij are the longitudinal and lateral offsets of the force arm between the electric wheel force and the center of mass, ΔF x,ij , ΔF y,ij are the change amounts of the longitudinal force and the lateral force of the tire, M t is the yaw moment caused by considering the asymmetry of the tire force, M cg is the yaw moment caused by the offset of the center of mass, d γ is the unobservable yaw moment disturbance, M dir is the yaw moment input by the driver, I z is the moment of inertia of the vehicle about the z-axis, γ is the yaw angular velocity, is the yaw angular acceleration; Design a low-frequency observer and use a wide time window to estimate the yaw moment M caused by the centroid offset cg ; Design a high-frequency observer for observation and use the residual of the low-frequency observer as the input of the high-frequency observer to estimate the yaw moment M caused by tire force asymmetry t ; The specific design of the yaw disturbance torque observer in step 22) is as follows: The specific design of the low-frequency disturbance observer is as follows: where t is the sampling time of the sensor; e1 is the estimated residual of the low-frequency disturbance observer, is the estimated value of the yaw rate; β 11 , β 12 is the gain coefficient of the low-frequency disturbance observer, balancing the convergence speed and noise resistance; is the estimated value of the yaw moment caused by the centroid offset; is the derivative with respect to the sensor sampling time; is the estimated value of the vehicle yaw rate ignoring the tire force asymmetry, is is the derivative with respect to the sensor sampling time; a x is the longitudinal acceleration of the vehicle; a y is the lateral acceleration of the vehicle; F z is the vertical force of the electric wheel; φ is the non-linear enhancement function, enhancing the tracking ability of the disturbance change; α is the parameter of the non-linear enhancement function; δ is the maximum estimated error of e1; the non-linear enhancement function φ is designed as: where λ is a parameter of the non - linear enhancement function, and the design of the parameter λ satisfies λ = k λ (1 - α), by adjusting the gain k λ to balance the fast response and smooth compensation; Use the estimated residual e1 in the low-frequency disturbance observer as the input to design the high-frequency disturbance observer. The specific design of the high-frequency disturbance observer is as follows: Among them, ω ij represents the rotational speed of each electric wheel, represents the change rate of the rotational speed of each electric wheel; the high-frequency disturbance observer error is the estimated value of e1; is the estimated value of the yaw angular velocity of the vehicle ignoring the centroid offset; is the estimated value of; β 21 , β 22 is the gain coefficient of the high-frequency disturbance observer; the nonlinear function ψ is designed as: where, δ2 is the maximum estimation error of e2; γ(e2) is the gain of the nonlinear function, and γ0 is the nominal gain value; The yaw disturbing moment finally obtained in step 23) is specifically expressed as follows:

5. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 4, characterized in that The specific steps for constructing a robust model predictive controller based on yaw disturbance torque compensation in step 3) are as follows: 31) Based on the lateral dynamic model of the commercial vehicle under uneven load distribution, establish a discretized state space equation; 32) Construct a two-degree-of-freedom reference model of the commercial vehicle to output the ideal yaw angular velocity and centroid sideslip angle as the tracking reference target of the controller; 33) On the basis of considering yaw disturbance torque compensation, design a nominal model predictive controller, and realize the prediction and control of the lateral state of the commercial vehicle by introducing a disturbance estimation compensation mechanism; 34) An auxiliary sliding mode controller is introduced into the nominal model predictive controller to enhance the anti-interference ability of the system through the design of robust control gain, and the coordinated yaw moment is output.

6. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 5, characterized in that, The discretized state space equation established in step 31) is specifically as follows: The lateral dynamic model of a commercial vehicle under the condition of uneven load distribution is described as: where C1, C2, and C3 are the cornering stiffnesses of the front axle tires, middle axle tires, and rear axle tires, respectively; a is the distance from the front axle to the center of gravity; b is the distance from the middle axle to the center of gravity; c is the distance between the middle axle and the rear axle; m is the vehicle mass; v is the vehicle speed; β is the sideslip angle of the vehicle center of mass, is the sideslip angular velocity of the vehicle center of mass; Define the state variable χ = [β, γ] T , and the input μ = M dir , the perturbation caused by the non-uniform load The system state equation is as follows: wherein, is the derivative of the state variable with respect to the sensor sampling time; the system matrix input matrix disturbance matrix Using a zero-order hold, setting the control input to be constant within the sampling period T, and discretizing the system state equation, the discretized state equation is obtained as follows: χ(k + 1)= A d χ(k)+ B d μ(k)+ D d ω(k) where χ(k) is the state variable at the current time k, and χ(k + 1) is the state variable at the next time; μ(k) is the input at the current time; ω(k) is the disturbance caused by the non-uniform load at the current time; A d = e AT is the discretized system matrix, is the discretized input matrix, is the discretized disturbance matrix, τ is the temporary time variable used in the integration process to represent the time-varying system state during the integration process; Step 32) specifically includes: The two-degree-of-freedom reference model of a commercial vehicle is specifically as follows: where v x is the longitudinal speed of the vehicle; δ f is the steering angle of the front wheels of the vehicle; The ideal yaw rate and sideslip angle of the center of mass output are specifically as follows: When the first derivative of the yaw rate is 0, the ideal yaw rate is controlled within the upper limit of the road adhesion coefficient, and the ideal yaw angular velocity γ d is expressed as: In the formula, μ is the road surface adhesion coefficient; Ideal centroidal sideslip angle β d = 0; The nominal model predictive controller in step 33) is specifically designed as follows: Use the yaw disturbing moment as the input compensation of the controller, and the compensation moment is designed as: where α * is the designed gain for adjusting the compensation amount; By compensating torque Reduce the yaw disturbing torque caused by uneven load The influence on vehicle dynamics, and restore the vehicle's motion to a relatively ideal state. The nominal model of the vehicle is defined as: where ξ(k) is the nominal state of the system; ξ(k + 1) is the nominal state of the system at the next time point; is the input of the nominal model, u(k) is obtained by optimizing and solving with the nominal model predictive controller; Predict the future state based on the nominal model state quantity of the vehicle at time k, and obtain: ξ(k) = Πξ(k) + ΞU(k) In the formula, ξ(k) is the future system nominal state matrix, which is used to represent the set of nominal states of the vehicle at multiple future times predicted from the current time k; the matrix Π is the state transition matrix, which is used to describe the state evolution law of the vehicle nominal model, indicating how the current state affects the future state; the matrix Ξ is the control gain matrix, which is used to describe the influence of the control quantity on the future state; the matrix U(k) is the control input matrix, which represents the sequence of control inputs applied at multiple future times starting from the current time k; Adopt a soft constraint method and set a cost function for optimal control to effectively ensure the tracking of the ideal yaw rate and sideslip angle of the center of mass; Definition χ r (k)=[β d γ d T , the cost function J(k) at time k in the nominal model predictive control is expressed as:​ where, N p is the state prediction interval, and N c is the control time domain; Q is the state weight matrix, R is the control input weight matrix, ε(k) is the slack constraint factor at time k, κ is the penalty coefficient, and i is the index variable of the predictive control step length; is the nominal state of the system at time k + i predicted at time k; is the nominal model input at time k + i predicted at time k; The constraint range of the nominal model should be shrunk under the constraint relationship of the actual system, expressed as: as follows: Among them, is the feasible region of the input of the reduced model control; φ * is the gain matrix of the influence of the disturbance input on the nominal system; η(k) is the perturbation boundary of the nominal model; u max and u min are the upper limit value and the lower limit value of the control input respectively; Expand and represent the objective function and combine the constraint conditions, and the form is as follows: In the formula, The optimal nominal control signal within the prediction range at the current time is obtained by solving the quadratic programming problem in the optimal control sequence, and the optimal control sequence is specifically as follows: Take the first term in the optimal control sequence U * (k) obtained by solving as the control signal of the nominal model, and it is required to satisfy the constraint relationship ​ The auxiliary sliding mode controller in step 34) is specifically designed as follows: Establish an error model to reflect the dynamic deviation between the actual system and the nominal model under the influence of external disturbances. The obtained state error equation is specifically: (k + 1) = A∈(k) + Bλ * η * (k) + ω * (k) Among them, represents the error state between the actual system and the nominal model, and ∈(k + 1) is the error state at the (k + 1)-th moment; λ * is the scale vector, and η * (k) is the auxiliary sliding mode control law to ensure the convergence of the error state, and ω * (k) is the unobservable disturbance remaining after compensation is completed; According to the error ∈(k) between the actual system and the nominal model, design the discrete switching function s(k) as: s(k) = C e ∈(k) Among them, is the sliding mode surface coefficient; The uncertainty perturbation ω * (k) in the state error equation is measured by using a delay estimation method, and the specific description is as follows: wherein, is the estimated value of the interference at the current moment; Design the discrete sliding mode reaching law as: Among them, α s , q s , t s , λ s , β s are the adjustment coefficients of the discrete sliding mode reaching law, satisfying 0 < α s < 1; 0 < q s t s < 1; 1 < λ s < β s ; σ(k) = ω(k) - ω(k - 1) represents the disturbance error within adjacent control periods; The auxiliary sliding mode control law η is calculated by combining the state error equation, the discrete switching function, and the discrete sliding mode reaching law * (k) is as follows: Within the sampling interval t, the total control output of the drive stability controller is specifically: In the formula, is the output signal of the driving stability controller, that is, the coordinated yaw moment to ensure the driving stability of the commercial vehicle.

7. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 6, characterized in that The specific steps of the driving force optimization distribution in step 4) are as follows: 41) Construct a vehicle friction circle based on the road surface adhesion load of a single electric wheel, and use it as the boundary of the feasible region to design a driving force optimization function; 42) Solve the driving force optimization function based on the constraints of the tire adhesion coefficient and the maximum output torque of the motor to satisfy the multi-axis coordinated yaw moment, and obtain the target driving force of each electric wheel.

8. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 7, characterized in that The driving force optimization function in step 41) is specifically designed as follows: The road adhesion load η of a single electric wheel ij It is represented by a friction ellipse: Among them, F x,ij is the longitudinal force of each electric wheel, F y,ij is the lateral force of each electric wheel, F z,ij is the vertical force of each electric wheel, and μ ij is the tire adhesion coefficient of each electric wheel; Considering the road surface adhesion load of a single electric wheel into the vehicle, the vehicle friction circle Φ is designed as follows: Among them, n is the number of electric wheels; The longitudinal force F of each electric wheel x,ij and the lateral force F of each electric wheel y,ij are related as follows: Among them, α ij is the tire slip angle of each electric wheel, C f,ij is the longitudinal stiffness of the tire, C r,ij is the lateral stiffness of the tire; Let Then the driving force optimization function J is specifically described as follows: Step 42) specifically includes: The constraints of the tire adhesion coefficient and the maximum output torque of the motor are specifically as follows: Among them, T max is the maximum torque that each electric wheel can output; R is the rolling radius of the electric wheel; The specific process of solving the target driving force of each electric wheel is as follows: While satisfying the constraints of tire adhesion coefficient and the maximum output torque of the motor, multi-axial coordinated yaw moment needs to be satisfied, specifically: Among them, F x,fl is the longitudinal force of the left front wheel, F x,fr is the longitudinal force of the right front wheel; F x,ml is the longitudinal force of the left middle wheel; F x,mr is the longitudinal force of the right middle wheel; F x,rl is the longitudinal force of the left rear wheel; F x,rr is the longitudinal force of the right rear wheel; B * is the wheelbase of the electric wheel; Use a sequential quadratic programming solver to solve the following equation to obtain the target driving forces of each electric wheel, as follows:

9. The commercial vehicle drive control method considering the uneven load distribution characteristics according to claim 8, characterized in that, The specific steps for designing the driving force synchronization tracking controller in step 5) are as follows: 51) Establish a tire relaxation characteristic model to describe the hysteresis characteristic of the driving force response; 52) Based on the target driving forces obtained in step 4), design a dynamic-steady decoupled dual reference model to generate a dynamic synchronization target value and a steady-state tracking target value; 53) Introduce a composite error based on the dynamic synchronization and steady-state tracking target values in step 52), and solve it by designing a driving force synchronization tracking controller based on the optimization of the hybrid energy field, and output the torque of the electric wheel that satisfies the driving force synchronization tracking control.

10. The commercial vehicle drive control method considering the characteristic of uneven load distribution according to claim 9, wherein, The specific construction of the tire relaxation characteristic model in step 51) is as follows: The tire relaxation longitudinal force equation is expressed as: where, σ ij represents the relaxation length of each electric wheel, i.e., the distance traveled by the vehicle from the start of tire deformation to reaching a steady state; the superscript L represents including the hysteresis effect; τ ij represents the time constant obtained by dividing the relaxation length of each electric wheel by the longitudinal speed of the vehicle; Based on the tire relaxation longitudinal force equation, the tire relaxation characteristic model is specifically as follows: where I ω is the moment of inertia of the electric wheel, T ij is the driving torque of the in-wheel motor, T f,ij is the rolling resistance torque, and R e is the effective rolling radius of the tire; The specifically designed dynamic-steady decoupled dual reference model in step 52) is as follows: Dynamic synchronization target values generated based on the tire relaxation model are as follows: In the formula, is a normalization factor used to map the dynamic response speeds of each electric wheel to a unified time scale, is the geometric mean relaxation time; is the target driving force obtained in step 4); Under steady-state conditions, the steady-state tracking target values of each electric wheel are defined as as follows: In the formula, is the reciprocal of the normalization factor; Step 53) is specifically as follows: Introduce a composite error based on the dynamic synchronization and steady-state tracking target values in step 52), as follows: In the formula, the dynamic error term is used to describe the actual force F of the electric wheel x,ij relative to the dynamic reference ; the steady-state error term is used to describe the actual force F of the electric wheel x,ij and the steady-state reference deviation; Based on the composite error, construct a hybrid energy function Ψ that includes synchronous potential energy, steady-state potential energy, and dissipated energy, which is defined as follows: where K d , K s , and K c are the dynamic stiffness coefficient, the steady-state stiffness coefficient, and the high-frequency suppression coefficient, respectively; Define the Lyapunov function based on the hybrid energy function Ψ as: Derivative with respect to V ij The derivative is obtained as follows: To ensure The torque of the electric wheel that satisfies the driving force synchronous tracking control is obtained as follows: where K v > 0 is the error adjustment gain, denotes the gradient direction of the mixed energy field function in the error space.

Images