Steering control method for distributed driving pilotless automobile

Abstract

The invention discloses a steering control method for a distributed driving unmanned automobile, and relates to the technical field of automobile steering control, and the method comprises the steps: obtaining operation state parameters and reference track information of the distributed driving unmanned automobile; based on the operation state parameters and the reference trajectory information, generating a target front wheel rotation angle by using an upper trajectory tracking controller, inputting the target front wheel rotation angle into an ideal two-degree-of-freedom model, and calculating to obtain target transverse parameters; based on the operation state parameters and the target transverse parameters, an additional yawing moment delta M is generated through a lower-layer four-wheel steering controller; and constructing an optimal torque distribution objective function, calculating the driving torque of each wheel by using a cuckoo algorithm when the optimal torque distribution objective function reaches the minimum, and performing steering control. According to the method, the transverse stability of the distributed driving unmanned automobile can be effectively improved, and safe and reliable running of the distributed driving unmanned automobile is guaranteed.

Description

Technical Field

[0001] The present application relates to the field of automobile steering control technology, and in particular to a steering control method for a distributed drive unmanned vehicle. Background Art

[0002] With the development of modern distributed autonomous vehicles (DAVs), increasing attention is being paid to their stability and safety. Four-Wheel Steering (4WS), with active input from both the front and rear wheels, improves the DAV's steering flexibility at low speeds and stability at high speeds. It also reduces the vehicle's slip angle during steering, thereby reducing the likelihood of sideslip and effectively improving path-following capability. This has become a trend in DAVs to improve their stability and safety. However, when faced with complex and steeply curved roads, excessive pursuit of trajectory tracking performance can lead to poor stability, posing a safety hazard and creating an uncomfortable ride for passengers. Therefore, improving the stability and safety of DAVs has become a major challenge in vehicle control system design.

[0003] Stability refers to the ability of a distributed-drive autonomous vehicle to travel along a reference trajectory through a four-wheel steering system, and to resist interference and maintain stability when disturbed by external factors. Stability during steering, in particular, is an important indicator for measuring the quality of a vehicle.

[0004] Currently, most steering control methods for distributed drive vehicles are based on traditional PID sliding mode control. This method offers simple control but slow dynamic response, is prone to overshoot, and suffers from poor control performance at high speeds. While traditional model predictive control methods offer significant advantages in dynamic control performance, their high computational complexity makes them inadequate for the real-time control applications of distributed drive autonomous vehicles. Alternatively, steering control can be achieved by distributing the drive torque across the four wheels, but most distributed drive vehicles employ either equal or vertical load distribution. These methods fail to achieve optimal wheel torque distribution, resulting in poor steering stability. Summary of the Invention

[0005] In response to the above-mentioned problems and technical requirements, this application proposes a steering control method for a distributed drive unmanned vehicle. The technical solution of this application is as follows:

[0006] A steering control method for a distributed drive unmanned vehicle comprises the following steps:

[0007] Obtaining operating state parameters and reference trajectory information of the distributed drive unmanned vehicle, where the operating state parameters include actual lateral parameters, which represent the actual lateral operating state of the distributed drive unmanned vehicle;

[0008] Based on the operating state parameters and reference trajectory information, the upper-level trajectory tracking controller is used to generate the target front wheel angle of the distributed drive autonomous vehicle. The upper-layer trajectory tracking controller is used to control the trajectory tracking performance of the distributed-drive autonomous vehicle. The higher the overlap between the driving trajectory of the distributed-drive autonomous vehicle and the reference trajectory, the better the trajectory tracking performance.

[0009] Set the target front wheel angle Input the ideal two-degree-of-freedom model of the distributed drive unmanned vehicle to calculate the target lateral parameters of the distributed drive unmanned vehicle; the ideal two-degree-of-freedom model indicates the target lateral parameters and target front wheel angle of the distributed drive unmanned vehicle. relationship;

[0010] Based on the operating state parameters and the target lateral parameters, the lower-level four-wheel steering controller is used to generate an additional yaw moment ΔM of the distributed drive unmanned vehicle. The lower-level four-wheel steering controller is used to control the lateral stability of the distributed drive unmanned vehicle. The smaller the difference between the actual lateral parameters of the distributed drive unmanned vehicle and the target lateral parameters, the higher the lateral stability.

[0011] An optimal torque distribution objective function is constructed according to the additional yaw moment ΔM, and the cuckoo algorithm is used to calculate the driving torque of each wheel of the distributed drive unmanned vehicle when the optimal torque distribution objective function is minimized. Steering control of the distributed drive unmanned vehicle is then performed according to the driving torque of each wheel.

[0012] The beneficial technical effects of this application are:

[0013] The present application discloses a steering control method for a distributed drive unmanned vehicle. The method is based on the layered sliding mode theory and achieves effective improvement of the lateral stability of the distributed drive unmanned vehicle while ensuring trajectory tracking performance by constructing an upper-layer trajectory tracking controller and a lower-layer four-wheel steering controller for coordinated control. This ensures that both trajectory tracking performance and lateral stability performance of the distributed drive unmanned vehicle can be taken into account during driving, thereby ensuring the safe and reliable driving of the distributed drive unmanned vehicle.

[0014] By designing a novel adaptive reaching law composed of sign, tangent, and linear functions, the system's velocity on the sliding surface is adaptively adjusted based on the state of the sliding surface, effectively reducing chatter. Furthermore, considering that the trajectory tracking of a distributed-drive autonomous vehicle is a second-order system, a hierarchical fast terminal sliding mode, suitable for second-order systems, is employed to construct the upper-level trajectory tracking controller and the lower-level lateral stability controller for the four-wheel steering controller. This effectively avoids the singularity issues that traditional fast terminal sliding modes introduce into second-order systems. Furthermore, a nonlinear sigmoid function is used to replace the original error variable in the hierarchical fast terminal sliding surface to introduce nonlinear terms, improving the system's ability to process nonlinear signals and ultimately achieving better control results.

[0015] The lateral stability controller of the underlying four-wheel steering controller controls the yaw stability of the distributed-drive autonomous vehicle. Taking into account the impact of the additional yaw moment on lateral stability, the additional yaw moment controller further controls the yaw stability of the distributed-drive autonomous vehicle. The dual guarantees of the lateral stability controller and the additional yaw moment controller ensure lateral stability control. Furthermore, a novel adaptive integral fast terminal sliding mode is used to replace the traditional PID sliding mode to construct the additional yaw moment controller, achieving precise control of the additional yaw moment. Four-wheel torque control is achieved through optimal allocation using a cuckoo algorithm, ensuring better lateral stability of the distributed-drive autonomous vehicle and adapting to emergency obstacle avoidance situations at both low and high speeds. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 is a flow chart of the steering control method.

[0017] Figure 2 The present invention is a block diagram of a control system for a distributed drive unmanned vehicle in an embodiment.

[0018] Figure 3 It is a two-degree-of-freedom single-track dynamic model of a distributed drive unmanned vehicle.

[0019] Figure 4 FIG. 1 is a schematic diagram of trajectory tracking control in an embodiment.

[0020] Figure 5 The figure is a flowchart of the optimal torque distribution of the cuckoo algorithm in one embodiment.

[0021] Figure 6 It is a trajectory tracking simulation waveform diagram under low-speed double lane-changing conditions in a simulation example.

[0022] Figure 7 It is a trajectory tracking simulation waveform diagram under high-speed double lane-changing conditions in a simulation example.

[0023] Figure 8 This is a simulation waveform of the actual yaw angular velocity of four-wheel steering control under low-speed double lane-changing conditions in a simulation example.

[0024] Figure 9 This is a simulation waveform of the actual yaw angular velocity of four-wheel steering control under high-speed double lane change conditions in a simulation example.

[0025] Figure 10 This is a simulation waveform of the actual center of mass sideslip angle during four-wheel steering control under low-speed double lane-changing conditions in a simulation example.

[0026] Figure 11 This is a simulation waveform of the actual center of mass sideslip angle during four-wheel steering control under high-speed double lane change conditions in a simulation example.

[0027] Figure 12 This is a simulation waveform diagram of the four-wheel torque during four-wheel steering control under low-speed double lane-changing conditions in a simulation example.

[0028] Figure 13 This is a four-wheel torque simulation waveform diagram of four-wheel steering control under high-speed double lane change conditions in a simulation example.

[0029] Figure 14 This is a simulation waveform diagram of the yaw stability of four-wheel steering control under low-speed double lane change conditions in a simulation example.

[0030] Figure 15 This is a simulation waveform diagram of the yaw stability of four-wheel steering control under high-speed double lane change conditions in a simulation example.

[0031] Figure 16 This is the control effect of different approach rates in a simulation example. DETAILED DESCRIPTION

[0032] The specific implementation of this application will be further described below with reference to the accompanying drawings.

[0033] This application discloses a distributed drive unmanned vehicle steering control method, please refer to Figure 1 The specific steps of the method are as follows:

[0034] Step 1: Use the onboard sensors to obtain the operating state parameters and reference trajectory information of the distributed drive unmanned vehicle during driving. The operating state parameters include actual lateral parameters, which represent the actual lateral operating state of the distributed drive unmanned vehicle.

[0035] Step 2: Based on the operating state parameters and reference trajectory information, the upper-level trajectory tracking controller is used to generate the target front wheel turning angle of the distributed drive autonomous vehicle. The upper-layer trajectory tracking controller is used to control the trajectory tracking performance of the distributed drive unmanned vehicle. The higher the overlap between the driving trajectory of the distributed drive unmanned vehicle and the reference trajectory, the better the trajectory tracking performance.

[0036] Since distributed drive unmanned vehicles need to travel along a preset reference trajectory, they need to closely track the reference trajectory during driving to prevent safety hazards caused by deviations from the driving trajectory of the distributed drive unmanned vehicles. In addition, while ensuring trajectory tracking performance, the stability of the vehicle must also be ensured, especially the lateral stability during steering. Therefore, controlling the safe and stable driving of distributed drive unmanned vehicles is a complex control problem. In order to achieve effective control of the entire system, based on the hierarchical sliding mode theory, the control problem of the entire system is processed in layers, and the complex large system is decomposed into two relatively simple subsystems at the upper and lower levels. Accurate and effective control is achieved through coordination and cooperation between the subsystems at each level. The control system of the distributed drive unmanned vehicle is constructed as follows: Figure 2 As shown in the figure, it includes an upper-level trajectory tracking controller, a lower-level four-wheel steering controller, a longitudinal vehicle speed controller, and an optimal torque distribution module. The upper-level trajectory tracking controller ensures trajectory tracking performance, the lower-level four-wheel steering controller ensures lateral stability, the longitudinal vehicle speed controller maintains a constant longitudinal vehicle speed, and the optimal torque distribution module achieves optimal four-wheel torque. The core of the entire control system lies in the coordinated cooperation between the upper-level trajectory tracking controller and the lower-level four-wheel steering controller to achieve both trajectory tracking performance and lateral stability. The longitudinal vehicle speed controller and the optimal torque distribution module assist the lower-level four-wheel steering controller to achieve more precise lateral stability control.

[0037] In one embodiment, a specific method for constructing an upper-layer trajectory tracking controller includes:

[0038] (1) Determine the trajectory tracking error e of the distributed drive unmanned vehicle based on the operating state parameters and reference trajectory information c .

[0039] The distributed drive autonomous vehicle in operation can be regarded as a complex nonlinear system, which often requires the construction of multiple differential equations of state quantities to represent the vehicle dynamics model. This application mainly studies the lateral dynamics of vehicle steering, so the model is simplified for easy control. A simplified two-degree-of-freedom single-track dynamics model is used, such as Figure 3 As shown in , the model ignores the effects of changes in suspension and tire characteristics and can simulate the stable driving state of the vehicle. When controlling the trajectory tracking of the distributed drive unmanned vehicle, the main focus is on observing the overlap between the actual driving trajectory and the reference trajectory, and adjusting the vehicle's operating state in real time based on the trajectory tracking error. Figure 4As shown in the figure, the operating state parameters of the distributed drive unmanned vehicle also include the actual heading angle φ, the actual longitudinal speed v x and actual lateral speed v y The actual lateral parameters include the actual yaw rate ω and the actual center of mass roll angle β, and the reference trajectory information includes the reference heading angle φ ref and trajectory curvature ρ, where the actual heading angle φ, the reference heading angle φ ref are the parameters in the geodetic coordinate system, Figure 4 The X and Y axes are the coordinate axes of the earth coordinate system; the remaining parameters are the parameters of the distributed drive unmanned vehicle body coordinate system. Please refer to the coordinate axes of the body coordinate system. Figure 3 The x-axis and y-axis in Figure 3 The CG in the figure is the center of mass of the distributed drive autonomous vehicle and the origin of the body coordinate system. The x-axis is along the vehicle's driving direction (longitudinal direction) and the y-axis is along the vehicle's lateral direction (transverse direction). In one embodiment, the trajectory tracking error e is calculated by tracking the vehicle's actual heading and actual position. c ,include:

[0040] According to the actual heading angle φ and the reference heading angle φ ref Determining the heading error Δ of a distributed drive autonomous vehicle φ ; According to the heading error Δ φ , actual longitudinal speed v x and actual lateral speed v y , determine the lateral error e of the distributed drive autonomous vehicle d ; The heading error and lateral error are further weighted to improve the control effect of the controller, and then the trajectory tracking error e of the distributed drive unmanned vehicle is determined c =m c1 e d +m c2 Δ φ , m c1 is the lateral error coefficient, m c2 is the heading error coefficient.

[0041] The heading error Δ in trajectory tracking of this application φ Defined as the difference between the actual heading angle φ and the reference heading angle φ ref The deviation of , expressed in the form of derivative is:

[0042]

[0043] Among them, ω ref represents the reference yaw rate, which is equal to the reference heading angle φ ref The derivative of The relationship between the reference heading angle and trajectory curvature of a distributed drive autonomous vehicle can be simplified as follows:

[0044]

[0045] like Figure 4 As shown, the lateral error e in trajectory tracking d It is defined as the lateral offset distance from the center of mass of the distributed drive autonomous vehicle to the reference trajectory, expressed in the form of a derivative:

[0046]

[0047] According to the small angle approximation, formula (1) and formula (3) can be simplified to:

[0048]

[0049] Derivative of formula (4) yields:

[0050]

[0051] right Figure 3 The two-degree-of-freedom single-track dynamic model shown in the figure is analyzed. When the distributed drive unmanned vehicle maintains a constant longitudinal speed, the actual two-degree-of-freedom model of the distributed drive unmanned vehicle is:

[0052]

[0053] in, is the derivative of the actual yaw rate, is the derivative of the actual center of mass sideslip angle, I z is the moment of inertia of the tire, a is the distance from the front wheel to the center of gravity of the distributed drive unmanned vehicle, b is the distance from the rear wheel to the center of gravity of the distributed drive unmanned vehicle, k1 is the lateral stiffness of the front wheel, k2 is the lateral stiffness of the rear wheel, m is the weight of the distributed drive unmanned vehicle, v x is the actual longitudinal speed, v y is the actual lateral speed, δ f is the actual front wheel turning angle, δ r is the actual rear wheel turning angle.

[0054] Take the actual state variable X = [ω β] T , the actual input variable u δ =[δ f ,δ r ] T , the actual output variable Rewritten into the form of state equation:

[0055]

[0056] Among them, the state variable coefficient matrix Input variable coefficient matrix Output variable coefficient matrix

[0057] The operating state of a front-wheel steering vehicle reflects the ideal vehicle state, that is, the vehicle will not suddenly produce excessive lateral acceleration or sideslip, and passengers will not feel shaking or discomfort. In order to ensure the comfort and stability of the vehicle during driving, keep the vehicle in a controllable and stable operating range, and avoid loss of control under extreme conditions, the actual lateral parameters of the vehicle are adjusted by controlling the actual front wheel angle to track the target lateral parameters. The target lateral parameters include the target yaw angular velocity ω d and the target center of mass sideslip angle β d , so let the actual rear wheel turning angle δ r = 0, simplifying the actual two-degree-of-freedom model to an ideal two-degree-of-freedom model:

[0058]

[0059] in, is the derivative of the ideal yaw rate, is the derivative of the ideal center of mass sideslip angle, is the target front wheel angle.

[0060] Since the longitudinal speed is constant, Heading error Δ φ The derivative and lateral error e d The derivatives of are derived separately, and combined with the ideal two-degree-of-freedom model, that is, the simultaneous solution of formulas (4), (5), and (8) to obtain the expression of the trajectory tracking error model:

[0061]

[0062] in, is the second derivative of the lateral error, is the second derivative of the heading error, is the derivative of the trajectory curvature.

[0063] (2) Based on the ideal two-degree-of-freedom model of the distributed drive autonomous vehicle, determine the target lateral parameters and target front wheel angle of the distributed drive autonomous vehicle The relationship between the trajectory tracking error e c The hierarchical fast terminal sliding surface of the upper trajectory tracking controller is designed; the hierarchical fast terminal sliding surface of the upper trajectory tracking controller includes an inner sliding surface s1 and an outer sliding surface s2.

[0064] Considering that trajectory tracking is a second-order system, it can be seen from formula (9) that the second-order derivatives of the lateral error and heading error are about the target front wheel angle The expression. On the numerator of the controller output solved by the traditional fast terminal sliding mode, there will be terms. Since both p and q are positive odd numbers and p < q, when the sliding mode surface s = 0, it will bring inevitable singularity problems to the system. Therefore, this application uses a hierarchical fast terminal sliding mode controller applicable to second-order systems to construct the upper-layer trajectory tracking controller.

[0065] In one embodiment, the hierarchical fast terminal sliding mode surface of the upper-layer trajectory tracking controller designed includes: based on the ideal two-degree-of-freedom model of a distributed-drive driverless vehicle, combined with the trajectory tracking error e c Determine the inner-layer sliding mode surface of the hierarchical fast terminal sliding mode surface Determine the outer-layer sliding mode surface of the hierarchical fast terminal sliding mode surface Among them, λ1, λ2 are inner-layer sliding mode surface coefficients greater than zero, a1, a2 are outer-layer sliding mode surface coefficients greater than zero, p and q are positive odd numbers, is the derivative of the trajectory tracking error, is the derivative of the inner-layer sliding mode surface.

[0066] (3) According to the hierarchical fast terminal sliding mode surface of the upper-layer trajectory tracking controller, and using a new type of adaptive reaching law, by making the switching point of the sign function adaptively advance in advance to constrain the movement of the system on the sliding mode surface so as to reduce chattering, determine the adaptive reaching law of the upper-layer trajectory tracking controller ε1, ε2, ε3 are reaching law coefficients greater than zero, e is the base of the natural logarithm, l0, ρ0 are regulation parameters greater than zero, s max is the maximum sliding mode distance of the upper-layer trajectory tracking controller, tan() is the tangent function, sign() is the sign function; among them, the reaching law coefficients and regulation parameters are all custom-set constants.

[0067] Prove the stability of the proposed upper-layer trajectory tracking controller based on the new type of adaptive reaching rate by analyzing the convergence time of the sliding mode surface. First, define the Lyapunov function And take the derivative of V to get:

[0068]

[0069] Assume that the initial state of the outer-layer sliding mode surface s2 is s t0 And s t0 >> 1, when s t0 ≥ s2 ≥ s t1 , where s t1 = s t0 -Δ, Δ is a relatively small positive number, the absolute value of the trajectory tracking error |e cWhen s>>0, the outer sliding surface is far from the stable state at this time, and the tangent function in formula (10) plays a major role. Therefore, the trajectory tracking error can converge to the sliding surface at a very fast speed, and the convergence time is t1:

[0070]

[0071] When s t1 >s2≥0, the outer sliding surface is closer to the stable state. The approaching rate of the tangent function is less than that of the linear function, and the sign function dominates. Especially when the absolute value of the trajectory tracking error |e c |→0, Therefore, the approaching time must be less than the approaching time of the reaching law of where The approaching time T of the finally designed adaptive reaching law s <t1+t2, and the outer sliding surface can converge in a finite time, that is, the inner sliding surface s1 converges to s1 = 0 within a finite time T s

[0072] Based on the hierarchical fast terminal sliding surface and adaptive reaching law of the upper-layer trajectory tracking controller, the upper-layer trajectory tracking controller is constructed.

[0073] Using the equivalent control principle, the trajectory tracking error model, the hierarchical fast terminal sliding surface and the adaptive reaching law of the upper-layer trajectory tracking controller are solved, and the output of the upper-layer trajectory tracking controller, that is, the target front wheel angle of the distributed drive driverless vehicle The expression of is:

[0074]

[0075] where, the expressions of the intermediate variables w1, w2, and w3 are:

[0076]

[0077] Step 3, input the target front wheel angle into the ideal two-degree-of-freedom model of the distributed drive driverless vehicle, and calculate the target lateral parameters of the distributed drive driverless vehicle; the ideal two-degree-of-freedom model indicates the relationship between the target lateral parameters of the distributed drive driverless vehicle and the target front wheel angle Substitute the target front wheel angle into the expression (8) of the ideal two-degree-of-freedom model, and the target yaw rate ω d and the target centroid side slip angle β d can be solved.

[0078] ​Step 4: Based on the operating state parameters and the target lateral parameters, the lower-level four-wheel steering controller is used to generate an additional yaw torque ΔM of the distributed drive unmanned vehicle. The lower-level four-wheel steering controller is used to control the lateral stability of the distributed drive unmanned vehicle. The smaller the difference between the actual lateral parameters of the distributed drive unmanned vehicle and the target lateral parameters, the higher the lateral stability.

[0079] Since the actual yaw rate ω and the actual sideslip angle β of the Braun-driven unmanned vehicle can reflect the lateral stability of the vehicle, the actual front wheel steering angle δ of the Braun-driven unmanned vehicle is controlled by the lower-level four-wheel steering controller. f and the actual rear wheel turning angle δ r , to adjust the actual yaw rate ω and the actual center of mass sideslip angle β to achieve the tracking target yaw rate ω d and the target center of mass sideslip angle β d Furthermore, the present application takes into account the effect of the additional yaw moment on lateral stability, and controls the additional yaw moment ΔM of the Braun-driven unmanned vehicle through the lower-level four-wheel steering controller to further adjust the actual yaw rate ω so that the vehicle operates at the target yaw rate ω. d .

[0080] In one embodiment, the lower four-wheel steering controller includes a lateral stability controller and an additional yaw moment controller. For the specific structure, please refer to Figure 2 The generation of the additional yaw moment ΔM using the lower-level four-wheel steering controller includes:

[0081] Based on the operating state parameters, target yaw rate ω d and the target center of mass sideslip angle β d , using the lateral stability controller to generate the actual front wheel steering angle δ f and the actual rear wheel turning angle δ r The lateral stability controller is used to control the yaw stability of the distributed drive unmanned vehicle. The actual yaw angular velocity ω is proportional to the target yaw angular velocity ω. d The smaller the difference between the actual center of mass sideslip angle β and the target center of mass sideslip angle β d The smaller the difference, the higher the yaw stability;

[0082] Based on the operating state parameters, the actual front wheel angle δ f and the actual rear wheel turning angle δ r , the additional yaw torque controller is used to generate the additional yaw torque ΔM of the distributed drive unmanned vehicle; the additional yaw torque controller is used to control the yaw stability of the distributed drive unmanned vehicle, and the actual yaw angular velocity ω is consistent with the target yaw angular velocity ω d The smaller the difference, the higher the yaw stability.

[0083] Since the lateral stability control of a distributed drive autonomous vehicle is also a second-order system, this application uses a hierarchical fast terminal sliding mode controller to construct a lateral stability controller. In one embodiment, constructing the lateral stability controller includes:

[0084] (1) According to the actual yaw rate ω and the actual sideslip angle β of the actual lateral parameters, the target yaw rate ω of the target lateral parameters d and the target center of mass sideslip angle β d , determine the lateral stability error E of the distributed drive unmanned vehicle, and E=[e1,e2] T , e1 is the yaw rate error and e1 = ω - ω d , e2 is the center of mass sideslip angle error and e2 = β-β d .

[0085] (2) Based on the actual two-degree-of-freedom model of the distributed drive unmanned vehicle, determine the actual yaw rate ω, the actual center of mass sideslip angle β, and the actual front wheel turning angle δ of the distributed drive unmanned vehicle. f and the actual rear wheel turning angle δ r The relationship between the lateral stability error E and the layered fast terminal sliding surface of the lateral stability controller is designed; the layered fast terminal sliding surface of the lateral stability controller includes the inner sliding surface S α =[S α1 ,S α2 ] T and the outer sliding surface S β =[S β1 ,S β2 ] T , S α1 、S α2 is the inner sliding mode variable, S β1 、S β2 is the outer sliding mode variable.

[0086] In order to solve the problem that the traditional PI sliding surface has poor ability to process nonlinear signals, this application replaces the error variable with a nonlinear sig function to obtain better control effect. In one embodiment, based on the actual two-degree-of-freedom model of the distributed drive unmanned vehicle, the inner sliding surface S of the layered fast terminal sliding surface is determined in combination with the lateral stability error E. α =[S α1 ,S α2 ] T =K p sig(E) τ +K i ∫sig(E) τ dt, determines the outer sliding surface of the layered fast terminal sliding surface

[0087] Among them, sig(E) τ =|E| τ sign(E), |E| τ Indicates that the diagonal elements are |e1| τ and |e2| τ The 2×2 main diagonal matrix, Indicates S α The matrix after each element of is subjected to p / q power operation, S αdt Indicates S α The matrix after the first-order derivative of each element of is obtained, p and q are positive odd numbers, τ is the power exponent set by the user; K p is the inner sliding surface coefficient matrix and kp1, kp2 are K p Matrix element value, K i is the inner sliding surface coefficient matrix and ki1 and ki2 are K i , A1 is the outer sliding surface coefficient matrix and a 11 、a 12 、a 21 、a 22 is the matrix element value of A1, A2 is the outer sliding surface coefficient matrix and b 11 、b 12 、b 21 、b 22 is the matrix element value of A2.

[0088] (3) According to the hierarchical fast terminal sliding surface of the lateral stability controller, the adaptive reaching law of the lateral stability controller is determined for:

[0089]

[0090] Among them, c1, c2, c3, d1, d2, d3 are reaching law coefficients greater than zero, l1, l2, ρ1, ρ2 are control parameters greater than zero, e is the base of natural logarithm, S max1 、S max2 is the maximum sliding mode distance of the lateral stability controller, [] T Represents matrix transpose;

[0091] Based on the hierarchical fast terminal sliding surface and adaptive reaching law of the lateral stability controller, a lateral stability controller is constructed.

[0092] The equivalent control principle is used to solve the actual two-degree-of-freedom model, the hierarchical fast terminal sliding surface and the adaptive reaching law of the lateral stability controller, and the output of the lateral stability controller is obtained, which is the actual front wheel turning angle δ of the distributed drive unmanned vehicle. f and the actual rear wheel turning angle δ r The expression is:

[0093]

[0094] where G is a 2×2 main diagonal matrix with τ as the diagonal element, |E| τ-1 Indicates that the diagonal elements are |e1 τ-1 | and |e2 τ-1 | is a 2×2 main diagonal matrix, X is the actual state variable and X=[ω,β] T , X d is the target state variable and X d =[ω d ,β d ] T , is the derivative of the target state variable, A is the state variable coefficient matrix of the state equation of the actual two-degree-of-freedom model and B is the input variable coefficient matrix of the state equation of the actual two-degree-of-freedom model and M is the integral form of the adaptive reaching law and () -1 Represents matrix inversion.

[0095] Since using only a lateral stability controller for lateral stability control is only applicable to conventional unmanned vehicles, and the speed of a distributed drive unmanned vehicle decreases during operation, it is necessary to control an additional yaw torque to further improve the control effect of tracking the ideal yaw angular velocity. In one embodiment, generating the additional yaw torque ΔM for the distributed drive unmanned vehicle using the additional yaw torque controller includes:

[0096] According to the actual yaw rate ω and the target yaw rate ω of the operating state parameters d , determine the yaw rate error e1 = ω-ω of the distributed drive unmanned vehicle d .

[0097] The additional yaw moment is introduced into the actual two-degree-of-freedom model of the distributed drive unmanned vehicle, and the additional two-degree-of-freedom model of the distributed drive unmanned vehicle is obtained as follows:

[0098]

[0099] According to formula (16), the additional yaw moment control of the distributed drive unmanned vehicle is a first-order system. Therefore, this application comprehensively considers the control effect and computational efficiency, and adopts an adaptive integral fast terminal sliding mode controller to construct an additional yaw moment controller to control the additional yaw moment of the vehicle so that the vehicle operates at the target yaw angular velocity. Based on the additional two-degree-of-freedom model, the actual yaw angular velocity ω and the actual center of mass sideslip angle β of the distributed drive unmanned vehicle are determined to be related to the additional yaw moment ΔM and the actual front wheel turning angle δ f and the actual rear wheel turning angle δ r The relationship between the yaw rate error e1 and the adaptive integral fast terminal sliding surface of the additional yaw torque controller is obtained. is an adaptive parameter and The derivative of The derivative of p and q are positive odd numbers, and ζ1 and ζ2 are constant coefficients.

[0100] Adaptive integral fast terminal sliding surface s based on the additional yaw moment controller ΔM , determine the adaptive reaching law of the additional yaw moment controller ε4, ε5, ε6 are reaching law coefficients greater than zero, l3, ρ3 are control parameters greater than zero, s maxΔM is the maximum sliding mode distance of the additional yaw moment controller. Select a suitable Lyapunov function and design the reaching law so that the sliding mode surface can eventually converge to 0. Due to the equivalent control principle Right now Convergence time of solution The yaw rate error e1 converges to 0 in a finite convergence time, indicating that the designed adaptive integral fast terminal sliding surface s ΔM Can make the system fast and stable.

[0101] The adaptive integral fast terminal sliding surface s of the additional two-degree-of-freedom model and the additional yaw moment controller is constructed using the equivalent control principle. ΔM and adaptive reaching law Solving the problem, we can obtain the output of the additional yaw moment controller, i.e., the additional yaw moment ΔM of the distributed drive unmanned vehicle, as:

[0102]

[0103] Step 5: construct an optimal torque distribution objective function based on the additional yaw moment ΔM, and use the cuckoo algorithm to calculate the driving torque of each wheel of the distributed drive unmanned vehicle when the optimal torque distribution objective function is minimized, and perform steering control on the distributed drive unmanned vehicle based on the driving torque of each wheel.

[0104] Based on the additional yaw moment ΔM obtained by the additional yaw moment controller, in order to achieve the lateral stability control effect required by the distributed drive autonomous vehicle, it is also necessary to use the optimal torque distribution module to obtain the optimal four-wheel torque. In one embodiment, determining the driving torque of each wheel of the distributed drive autonomous vehicle includes:

[0105] Since the speed of the distributed drive unmanned vehicle will decrease during operation, in order to ensure the constant longitudinal speed, an external drive is required. Based on the operating state parameters, this application uses the longitudinal speed controller to determine the total driving torque F to control the longitudinal speed to remain constant. x , where the longitudinal vehicle speed controller adopts PID controller.

[0106] By distributing the longitudinal driving force of the wheels, the tire utilization rate is minimized, thereby increasing the tire lateral force margin and improving the stability of the vehicle. At the same time, the optimal torque distribution objective function is established by combining the minimum road adhesion coefficient. F xi 、F zi (i=1, 2, 3, 4) represent the longitudinal driving force and vertical load of the left front wheel, right front wheel, left rear wheel, and right rear wheel respectively.

[0107] Construct the additional yaw moment ΔM required to maintain lateral stability and the total driving torque F required to maintain vehicle speed x The constraint matrix is: Bu = v. Construct the maximum output torque T of the wheel subjected to the hub motor. max and the inequality constraints limited by the road adhesion condition: r i is the tire radius.

[0108] The standard form of quadratic programming is: minJ = u T Wu Considering that the solution of Bu=v within the constraint range may have a unique solution, no solution, or infinite solutions, the equality constraint Bu=v is transformed into min||Bu=v||2 and used as part of the optimization objective to minimize the allocation error, forming a sequential least squares problem:

[0109]

[0110] in, u They represent the upper and lower limits of u, which can be determined according to the inequality constraints.

[0111] The weight coefficient γ is introduced to transform the sequential least squares problem into a weighted least squares problem. The optimal torque distribution objective function is constructed as follows:

[0112]

[0113] Among them, γ is the weight coefficient, and the control vector u=[F x1 F x2 F x3 F x4 ] T , input vector v=[F x ΔM], F x1 is the longitudinal driving force of the left front wheel, F x2 is the longitudinal driving force of the left rear wheel, F x3 is the longitudinal driving force of the right front wheel, F x4 is the longitudinal driving force of the right rear wheel; the coefficient matrix The weight matrix of the control vector u The weight matrix W of the input vector v v =diag(w Tx ,w ΔM ), diag() represents the main diagonal matrix, w Tx is the driving torque weight, w ΔM is the additional yaw moment weight, c i is the longitudinal driving force weight of the wheel, F zi is the vertical load on the wheel, μ i is the friction coefficient between the wheel and the road, and the integer parameters i = 1, 2, 3, 4 represent the left front wheel, right front wheel, left rear wheel, and right rear wheel respectively;

[0114] The cuckoo algorithm is combined with the Levy flight-and-walk mechanism to determine the weight coefficient γ that minimizes the actual yaw rate. For the optimization process of the cuckoo algorithm, please refer to Figure 5 The actual yaw angular velocity is used as the Cuckoo Algorithm's fitness value. The Cuckoo Algorithm also improves the algorithm's randomness by randomly generating a uniformly distributed value R between 0 and 1 during each iteration, as well as the probability Pa of the nest owner discovering an alien egg. This prevents the algorithm from falling into a local optimum. The specific implementation of the Cuckoo Algorithm and the Levy Flight and Walk mechanism is described in the prior art and will not be elaborated upon in this application.

[0115] After determining the global optimal weight coefficient γ, substitute the weight coefficient γ into the optimal torque distribution objective function, and use the effective set algorithm to solve the longitudinal driving force of each wheel when the optimal torque distribution objective function is minimized. xi and the tire radius r of each wheel i Calculate the driving torque T of each wheel i =F xi ·r i The effective set algorithm is a conventional method for solving weighted least squares problems, and its details will not be described in detail here.

[0116] In order to verify the control effect of the method of the present application, simulation control experiments were carried out on the vehicle conditions under low-speed and high-speed double lane-changing conditions. Figure 6 、 Figure 7 The comparison results of trajectory tracking simulation waveforms of the traditional method and the method of this application under low-speed and high-speed double-lane-changing conditions are shown. The traditional AFS strategy based on the driver model has the largest lateral deviation and average deviation at low speed. The AFS strategy that comes with carsim also has a large following error. The trajectory tracking controller designed with PID sliding mode can effectively reduce the lateral deviation peak and average deviation, but it will cause trajectory overshoot. The upper-layer trajectory tracking controller designed with layered sliding mode in this application can further reduce the lateral deviation peak and average deviation while effectively suppressing overshoot. The lateral deviation peak is optimized by 78.44%, 81.94% and 69.13% respectively compared with the previous three strategies, and the average lateral deviation is optimized by 86.5%, 91.52% and 82.3% respectively compared with the previous three strategies. 2%; the lateral deviation and average deviation of the trajectory tracking strategy using PID sliding mode design at high speed are the largest, and the layered sliding mode can reduce them by 19.03% and 34.43% respectively. The AFS strategy that comes with carsim and the layered sliding mode strategy of this study can both effectively suppress the average lateral deviation. Although both carsimAFS and AFS strategies use the driver's preview principle and can turn in advance to reduce the lateral deviation peak, they will produce a large trajectory overshoot after the double-lane trajectory tracking is completed, just like the trajectory tracking strategy using PID sliding mode design. However, the layered sliding mode strategy of this application can effectively suppress the trajectory overshoot after the trajectory tracking is completed, ensuring the accuracy of trajectory tracking.

[0117] Figure 8 、 Figure 9 The actual yaw rate simulation waveforms and yaw rate error comparison results of the traditional method and the method of the present application for four-wheel steering control under low-speed and high-speed double lane change conditions are demonstrated. Figure 10 、 Figure 11The comparison results of the actual center of mass slip angle simulation waveforms of the traditional method and the method of the present application for four-wheel steering control under low-speed and high-speed double-lane-shifting conditions are shown. When the layered sliding mode rear-wheel steering control strategy is adopted at low speed, the peak value of the yaw rate error is small, and the peak value of the center of mass slip angle error is the largest. When the traditional PID sliding mode four-wheel steering is adopted, the yaw rate error peak value increases by 90.55% compared with the layered sliding mode rear-wheel steering, but the center of mass slip angle error is greatly improved, and is reduced by 97.35% compared with the layered sliding mode rear-wheel steering; when the new layered fast terminal sliding mode control of the present application is adopted, the yaw rate error is only 0.002, which is reduced by 68.5% compared with the layered sliding mode rear-wheel steering, and the center of mass slip angle error peak value is only 0.00053deg, which is reduced by 98.47% compared with the layered sliding mode rear-wheel steering and 42.39% compared with the traditional PID sliding mode four-wheel steering, indicating that the new layered fast terminal sliding mode control of the present application is effective. The fast terminal sliding mode (HFTSM) control has good lateral stability control performance at low speeds. When the layered sliding mode rear-wheel steering control strategy is adopted at high speeds, the peak and average values ​​of the yaw rate error are the largest, and the peak and average values ​​of the center of mass slip angle error are the largest. When the traditional PID sliding mode four-wheel steering is adopted, the peak and average values ​​of the yaw rate error are improved to a certain extent, and the average value of the center of mass slip angle error is reduced by 92.548%. When the new layered fast terminal sliding mode control of this application is adopted, the effect is the best. The peak value of the yaw rate error is optimized by 54.7% and 38.16% compared with the previous two strategies, the average value of the yaw rate error is optimized by 19.98% and 14.35% compared with the previous two strategies, and the peak and average value of the center of mass slip angle error are reduced by 96.9% and 58.46%. This shows that the new layered fast terminal sliding mode control of this application has good lateral stability control performance at high speeds.

[0118] Figure 12 、 Figure 13 The comparison results of four-wheel torque simulation waveforms between the traditional method and the method of this application under the optimal distribution of additional yaw torque based on the new adaptive integral fast terminal sliding mode (AIFTSM) and cuckoo algorithm (CS) at low and high speeds are shown. Figure 14 、 Figure 15Comparisons of yaw stability simulation waveforms at low and high speeds between conventional methods and the present invention's method using a novel adaptive integral fast terminal sliding mode and cuckoo algorithm for optimal allocation of additional yaw torque are presented. At low speeds, the peak and average yaw rate errors are highest when the average allocation strategy for the additional yaw torque is employed, followed by the vertical load allocation strategy. However, when the present invention's optimal allocation strategy, based on the cuckoo algorithm, is employed, the peak yaw rate error is optimized by 58.63% and 40.59%, respectively, while the average yaw rate error is optimized by 67.62% and 46.12%, respectively, compared to the first two strategies. This demonstrates that the present invention's optimal allocation strategy, based on the cuckoo algorithm, offers superior control over lateral stability at low speeds. By replacing the traditional PID sliding mode with the novel adaptive integral sliding mode of this application, the peak yaw rate error was reduced by 22.84% and the average yaw rate error by 9.71%, further improving lateral stability and suppressing error chattering caused by the PID sliding mode. At high speeds, the average yaw rate error was greatest when the average distribution strategy for the additional yaw torque was employed, followed by the vertical load distribution strategy. However, when the optimal distribution strategy based on the cuckoo algorithm of this application was employed, the average yaw rate error was improved by 48.64% and 30.26%, respectively. This demonstrates that the optimal distribution strategy based on the cuckoo algorithm of this application can effectively suppress the overall yaw rate error at high speeds. Regardless of the distribution strategy employed, a large peak yaw rate error will occur. However, by replacing the traditional PID sliding mode with the novel adaptive integral fast terminal sliding mode of this application, the peak yaw rate error was effectively suppressed, reducing it by 20.96%.

[0119] Figure 16 The simulation comparison results of the approach rate control effect of using the tanh function instead of the sign function to optimize the approach rate, the adaptive improvement of the approach rate and the method of this application are shown. Figure 16 (a) It can be seen that when the double lane change working condition steering is completed, the actual front wheel angle of the tanh function optimization reaching law stabilizes faster and the angle does not overshoot, but due to Figure 16 (b) It can be seen that the actual front wheel steering angle jitter is larger when the tanh function-optimized convergence law is used. When the existing adaptive convergence law is used, the steering angle jitter is significantly suppressed, but the steering angle stabilization time and overshoot increase. When the new adaptive convergence law of the present application is used, the rapidity is further improved compared to the convergence law optimized by the tanh function, and the overshoot is only 0.008 degrees, which is reduced by 78.38% compared to the existing adaptive convergence law. At the same time, the jitter during stabilization is only 0.0042 degrees, which is reduced by 69.78% and 50% respectively compared to the first two convergence laws, which is well suppressed and improved.

[0120] The above description is only a preferred embodiment of the present application, and the present application is not limited to the above embodiments. It is understood that other improvements and variations directly derived or imagined by those skilled in the art without departing from the spirit and concept of the present application should be considered to be included in the scope of protection of the present application.

Claims

1. A steering control method for a distributed drive unmanned vehicle, characterized in that: The steering control method comprises: Obtaining operating state parameters and reference trajectory information of the distributed drive unmanned vehicle, wherein the operating state parameters include actual lateral parameters, which represent an actual lateral operating state of the distributed drive unmanned vehicle; Based on the operating state parameters and the reference trajectory information, the upper trajectory tracking controller is used to generate the target front wheel turning angle of the distributed drive unmanned vehicle. The upper-layer trajectory tracking controller is used to control the trajectory tracking performance of the distributed-drive unmanned vehicle. The higher the overlap between the driving trajectory of the distributed-drive unmanned vehicle and the reference trajectory, the better the trajectory tracking performance. The target front wheel angle The ideal two-degree-of-freedom model of the distributed drive unmanned vehicle is input to calculate the target lateral parameters of the distributed drive unmanned vehicle; the ideal two-degree-of-freedom model indicates the target lateral parameters of the distributed drive unmanned vehicle and the target front wheel turning angle. relationship; generating an additional yaw moment ΔM of the distributed drive unmanned vehicle using a lower-layer four-wheel steering controller based on the operating state parameter and the target lateral parameter; the lower-layer four-wheel steering controller is used to control the lateral stability of the distributed drive unmanned vehicle, and the smaller the difference between the actual lateral parameter and the target lateral parameter of the distributed drive unmanned vehicle, the higher the lateral stability; An optimal torque distribution objective function is constructed based on the additional yaw moment ΔM, and a cuckoo algorithm is used to calculate the driving torque of each wheel of the distributed drive unmanned vehicle that minimizes the optimal torque distribution objective function. Steering control of the distributed drive unmanned vehicle is performed based on the driving torque of each wheel.

2. The steering control method according to claim 1, characterized in that: Constructing the upper-layer trajectory tracking controller includes: Determine the trajectory tracking error e of the distributed drive unmanned vehicle based on the operating state parameters and the reference trajectory information. c ; Determine the target lateral parameters and target front wheel angle of the distributed drive autonomous vehicle based on the ideal two-degree-of-freedom model of the distributed drive autonomous vehicle The relationship between the trajectory tracking error e c The hierarchical fast terminal sliding surface of the upper trajectory tracking controller is designed; the hierarchical fast terminal sliding surface of the upper trajectory tracking controller includes an inner sliding surface s1 and an outer sliding surface s2; According to the hierarchical fast terminal sliding surface of the upper trajectory tracking controller, the adaptive reaching law of the upper trajectory tracking controller is determined ε1, ε2, ε3 are reaching law coefficients greater than zero, l0, ρ0 are control parameters greater than zero, e is the base of natural logarithm, s max is the maximum sliding mode distance of the upper trajectory tracking controller, tan() is the tangent function, and sign() is the sign function; The upper-layer trajectory tracking controller is constructed based on a hierarchical fast terminal sliding mode surface and an adaptive reaching law of the upper-layer trajectory tracking controller.

3. The steering control method according to claim 2, characterized in that: The operating state parameters also include actual heading angle φ, actual longitudinal speed v x and actual lateral speed v y , the actual lateral parameters include the actual yaw rate ω, and the reference trajectory information includes the reference heading angle φ ref and trajectory curvature ρ, the hierarchical fast terminal sliding surface of the upper trajectory tracking controller is designed to include: According to the actual heading angle φ and the reference heading angle φ ref Determining the heading error Δ of a distributed drive autonomous vehicle φ And the derivative of the heading error According to the heading error Δ φ , actual longitudinal speed v x and actual lateral speed v y , determine the lateral error e of the distributed drive autonomous vehicle d And the derivative of the lateral error Determining the trajectory tracking error e of a distributed-drive autonomous vehicle c =m c1 e d +m c2 Δ φ , m c1 is the lateral error coefficient, m c2 is the heading error coefficient; Based on the ideal two-degree-of-freedom model of distributed drive unmanned vehicles, combined with the trajectory tracking error e c Determine the inner sliding surface of the layered rapid terminal sliding surface Determine the outer sliding surface of the layered rapid terminal sliding surface Among them, λ1 and λ2 are the inner sliding surface coefficients greater than zero, a1 and a2 are the outer sliding surface coefficients greater than zero, p and q are positive odd numbers, is the derivative of the trajectory tracking error, is the derivative of the inner sliding surface.

4. The steering control method according to claim 3, characterized in that: The target lateral parameters include the target yaw rate ω d and the target center of mass sideslip angle β d , generating the target front wheel steering angle of the distributed drive autonomous vehicle include: The ideal two-degree-of-freedom model of a distributed drive autonomous vehicle is: in, is the derivative of the ideal yaw rate, is the derivative of the ideal center of mass sideslip angle, I z is the moment of inertia of the tire, a is the distance from the front wheel to the center of gravity of the distributed drive unmanned vehicle, b is the distance from the rear wheel to the center of gravity of the distributed drive unmanned vehicle, k1 is the lateral stiffness of the front wheel, k2 is the lateral stiffness of the rear wheel, m is the weight of the distributed drive unmanned vehicle, v x is the actual longitudinal speed, v y is the actual lateral vehicle speed; The heading error Δ φ The derivative and lateral error e d The derivatives of are derived separately, and combined with the ideal two-degree-of-freedom model, the expression of the trajectory tracking error model is obtained as follows: in, is the second derivative of the lateral error, is the second derivative of the heading error, is the derivative of the trajectory curvature; The trajectory tracking error model, the hierarchical fast terminal sliding surface and the adaptive reaching law of the upper trajectory tracking controller are solved using the equivalent control principle to obtain the target front wheel turning angle of the distributed drive unmanned vehicle. The expression is: Among them, the expressions of intermediate variables w1, w2, and w3 are:

5. The steering control method according to claim 1, characterized in that: The actual lateral parameters include the actual yaw rate ω and the actual center of mass sideslip angle β, and the target lateral parameters include the target yaw rate ω d and the target center of mass sideslip angle β d The lower four-wheel steering controller includes a lateral stability controller and an additional yaw moment controller, and the use of the lower four-wheel steering controller to generate an additional yaw moment ΔM includes: Based on the operating state parameters, the target yaw angular velocity ω d and the target center of mass sideslip angle β d , using the lateral stability controller to generate the actual front wheel steering angle δ f and the actual rear wheel turning angle δ r The lateral stability controller is used to control the yaw stability of the distributed drive unmanned vehicle, and the actual yaw angular velocity ω is consistent with the target yaw angular velocity ω. d The smaller the difference between the actual center of mass sideslip angle β and the target center of mass sideslip angle β d The smaller the difference, the higher the yaw stability; Based on the operating state parameters, the actual front wheel turning angle δ f and the actual rear wheel turning angle δ r , an additional yaw torque controller is used to generate an additional yaw torque ΔM of the distributed drive unmanned vehicle; the additional yaw torque controller is used to control the yaw stability of the distributed drive unmanned vehicle, and the actual yaw angular velocity ω is consistent with the target yaw angular velocity ω d The smaller the difference, the higher the yaw stability.

6. The steering control method according to claim 5, characterized in that: Building a lateral stability controller involves: According to the actual yaw rate ω and the actual sideslip angle β of the actual lateral parameter, the target yaw rate ω of the target lateral parameter d and the target center of mass sideslip angle β d , determine the lateral stability error E of the distributed drive unmanned vehicle; Based on the actual two-degree-of-freedom model of the distributed drive unmanned vehicle, the actual yaw rate ω, the actual center of mass sideslip angle β and the actual front wheel turning angle δ of the distributed drive unmanned vehicle are determined. f and the actual rear wheel turning angle δ r The relationship between the lateral stability error E and the layered fast terminal sliding mode surface of the lateral stability controller is designed; the layered fast terminal sliding mode surface of the lateral stability controller includes the inner sliding mode surface S α =[S α1 ,S α2 ] T and the outer sliding surface S β =[S β1 ,S β2 ] T , S α1 、S α2 is the inner sliding mode variable, S β1 、S β2 is the outer sliding mode variable; According to the hierarchical fast terminal sliding surface of the lateral stability controller, the adaptive reaching law of the lateral stability controller is determined. for: Among them, c1, c2, c3, d1, d2, d3 are reaching law coefficients greater than zero, l1, l2, ρ1, ρ2 are control parameters greater than zero, e is the base of natural logarithm, S max1 、S max2 is the maximum sliding mode distance of the lateral stability controller, tan() is the tangent function, sign() is the sign function, [] T Represents matrix transpose; The lateral stability controller is constructed based on a hierarchical fast terminal sliding mode surface and an adaptive reaching law of the lateral stability controller.

7. The steering control method according to claim 6, characterized in that: The hierarchical fast terminal sliding surface of the designed lateral stability controller includes: Determine the lateral stability error E = [e1, e2] of the distributed drive autonomous vehicle T , e1 is the yaw rate error and e1 = ω - ω d , e2 is the center of mass sideslip angle error and e2 = β-β d ; Based on the actual two-degree-of-freedom model of the distributed drive unmanned vehicle, the inner sliding surface S of the layered fast terminal sliding surface is determined in combination with the lateral stability error E. α =[S α1 ,S α2 ] T =K p sig(E) τ +K i ∫sig(E) τ dt, determines the outer sliding surface of the layered fast terminal sliding surface Among them, sig(E) τ =|E| τ sign(E), |E| τ Indicates that the diagonal elements are |e1| τ and |e2| τ The main diagonal matrix of , τ is the power index, Indicates S α The matrix after each element of is subjected to p / q power operation, S αdt Indicates S α The matrix after the first-order derivative of each element of is obtained, p and q are positive odd numbers; K p is the inner sliding surface coefficient matrix and kp1, kp2 are K p Matrix element value, K i is the inner sliding surface coefficient matrix and ki1 and ki2 are K i , A1 is the outer sliding surface coefficient matrix and a 11 、a 12 、a 21 、a 22 is the matrix element value of A1, A2 is the outer sliding surface coefficient matrix and b 11 、b 12 、b 21 、b 22 is the matrix element value of A2.

8. The steering control method according to claim 7, characterized in that: Generate actual front wheel steering angle δ f and the actual rear wheel turning angle δ r include: The actual two-degree-of-freedom model of a distributed drive autonomous vehicle is: in, is the derivative of the actual yaw rate, is the derivative of the actual center of mass sideslip angle, I z is the moment of inertia of the tire, a is the distance from the front wheel to the center of gravity of the distributed drive unmanned vehicle, b is the distance from the rear wheel to the center of gravity of the distributed drive unmanned vehicle, k1 is the lateral stiffness of the front wheel, k2 is the lateral stiffness of the rear wheel, m is the weight of the distributed drive unmanned vehicle, v x is the actual longitudinal speed, v y is the actual lateral vehicle speed; The actual two-degree-of-freedom model, the hierarchical fast terminal sliding surface and the adaptive reaching law of the lateral stability controller are solved using the equivalent control principle to obtain the actual front wheel turning angle δ of the distributed drive unmanned vehicle. f and the actual rear wheel turning angle δ r The expression is: where G is the main diagonal matrix with diagonal elements τ, |E| τ-1 Indicates that the diagonal elements are |e1 τ-1 | and |e2 τ-1 |, X is the actual state variable and X = [ω, β] T , X d is the target state variable and X d =[ω d ,β d ] T , is the derivative of the target state variable, A is the state variable coefficient matrix of the state equation of the actual two-degree-of-freedom model and B is the input variable coefficient matrix of the state equation of the actual two-degree-of-freedom model and M is the integral form of the adaptive reaching law and () -1 Represents matrix inversion.

9. The steering control method according to claim 5, characterized in that: Generating the additional yaw moment ΔM of the distributed drive autonomous vehicle by using the additional yaw moment controller includes: According to the actual yaw rate ω and the target yaw rate ω of the actual lateral parameter d , determine the yaw rate error e1 = ω-ω of the distributed drive unmanned vehicle d ; The additional yaw moment is introduced into the actual two-degree-of-freedom model of the distributed drive unmanned vehicle, and the additional two-degree-of-freedom model of the distributed drive unmanned vehicle is obtained as follows: in, is the derivative of the actual yaw rate, is the derivative of the actual center of mass sideslip angle, ΔM is the additional yaw moment, I z is the moment of inertia of the tire, a is the distance from the front wheel to the center of gravity of the distributed drive unmanned vehicle, b is the distance from the rear wheel to the center of gravity of the distributed drive unmanned vehicle, k1 is the lateral stiffness of the front wheel, k2 is the lateral stiffness of the rear wheel, m is the weight of the distributed drive unmanned vehicle, and v is x Actual longitudinal speed, v y is the actual lateral vehicle speed; Based on the additional two-degree-of-freedom model, the actual yaw rate ω and the actual center of mass sideslip angle β of the distributed drive unmanned vehicle are determined along with the additional yaw moment and the actual front wheel turning angle δ. f and the actual rear wheel turning angle δ r The relationship between the yaw rate error e1 and the adaptive integral fast terminal sliding mode surface of the additional yaw torque controller is obtained. is an adaptive parameter and The derivative of The derivative of p and q are positive odd numbers, ζ1 and ζ2 are constant coefficients; Adaptive integral fast terminal sliding surface s based on the additional yaw moment controller ΔM , determine the adaptive reaching law of the additional yaw moment controller ε4, ε5, ε6 are reaching law coefficients greater than zero, l3, ρ3 are control parameters greater than zero, s maxΔM is the maximum sliding mode distance of the additional yaw moment controller, tan() is the tangent function, and sign() is the sign function; The adaptive integral fast terminal sliding surface s of the additional two-degree-of-freedom model and the additional yaw moment controller is adjusted by using the equivalent control principle. ΔM and adaptive reaching law Solving the problem, we can obtain the additional yaw moment ΔM of the distributed drive unmanned vehicle as: Among them, I z is the tire moment of inertia, a is the distance from the front wheel to the center of gravity of the distributed drive unmanned vehicle, b is the distance from the rear wheel to the center of gravity of the distributed drive unmanned vehicle, k1 is the front wheel cornering stiffness, k2 is the rear wheel cornering stiffness, v x is the actual longitudinal speed, v y is the actual lateral speed, is the derivative of the ideal yaw rate.

10. The steering control method according to claim 1, characterized in that: Determining the driving torque of each wheel of a distributed drive autonomous vehicle includes: Based on the operating state parameters, the longitudinal vehicle speed controller is used to determine the total driving torque F that controls the longitudinal vehicle speed to remain constant. x , according to the additional yaw moment ΔM and the total drive torque F x The constructed optimal torque distribution objective function is: Among them, γ is the weight coefficient, and the control vector u=[F x1 F x2 F x3 F x4 ] T , input vector v=[F x ΔM], F x1 is the longitudinal driving force of the left front wheel, F x2 is the longitudinal driving force of the left rear wheel, F x3 is the longitudinal driving force of the right front wheel, F x4 is the longitudinal driving force of the right rear wheel; the coefficient matrix The weight matrix of the control vector u The weight matrix W of the input vector v v =diag(w Tx ,w ΔM ), diag() represents the main diagonal matrix, w Tx is the driving force weight, w ΔM is the additional yaw moment weight, c i is the longitudinal driving torque weight of the wheel, F zi is the vertical load on the wheel, μ i is the friction coefficient between the wheel and the road, and the integer parameters i = 1, 2, 3, 4 represent the left front wheel, right front wheel, left rear wheel, and right rear wheel respectively; The cuckoo algorithm is combined with the Levy flight-and-wander mechanism to determine the weight coefficient γ that minimizes the actual yaw rate. The weight coefficient γ is substituted into the optimal torque distribution objective function, and the effective set algorithm is used to solve the longitudinal driving force of each wheel when the optimal torque distribution objective function is minimized. The driving torque of each wheel is calculated based on the longitudinal driving force of each wheel and the tire radius of each wheel.

Images