Distributed drive intelligent vehicle control method based on longitudinal and lateral coupling model prediction
By establishing a distributed drive intelligent vehicle control method based on a horizontal and vertical coupling model prediction, the problem of vehicle instability under extreme conditions caused by traditional control methods is solved, achieving high-precision trajectory tracking and stable control, and improving vehicle performance under complex conditions.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- EAST CHINA JIAOTONG UNIVERSITY
- Filing Date
- 2026-03-31
- Publication Date
- 2026-07-07
Smart Images
Figure CN121947531B_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent vehicle technology, and more specifically to a distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction. Background Technology
[0002] With the development of advanced autonomous driving technology, vehicles need to achieve high-precision trajectory tracking under a wider range of operating conditions. Trajectory tracking control mainly focuses on the vehicle's ability to follow the desired path, typically achieved by adjusting the front wheel steering angle and longitudinal acceleration. However, under extreme or dynamic conditions, simple trajectory tracking control may drive the vehicle close to or beyond its adhesion limits, inducing instability such as oversteer or understeer.
[0003] Existing technologies mostly employ a hierarchical control architecture: an upper-level trajectory planner generates a reference trajectory, middle-level lateral and longitudinal controllers independently calculate control inputs, and lower-level actuators provide control. This architecture decouples lateral and longitudinal control, simplifying the design, but it has shortcomings when dealing with dynamic scenarios where lateral and longitudinal dynamics are strongly coupled. On the one hand, independent lateral and longitudinal control objectives may conflict; for example, steering commands applied to quickly track a curve trajectory may lead to a decrease in trajectory tracking accuracy. On the other hand, traditional model predictive control typically uses linear models with fixed parameters; when vehicle parameters change over time or operating conditions change drastically, model mismatch can lead to decreased control performance or even instability. Summary of the Invention
[0004] The purpose of this invention is to provide a distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction, so as to improve trajectory tracking accuracy and vehicle control performance.
[0005] A distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction includes:
[0006] Step S1: Acquire multi-source vehicle data of the distributed drive intelligent vehicle through on-board sensors. The multi-source vehicle data includes vehicle dynamics data and road environment data.
[0007] Step S2: Based on multi-source vehicle data, establish a lateral and longitudinal coupled vehicle state-space equation that considers yaw.
[0008] Step S3: Based on the lateral and longitudinal coupled vehicle state space equation considering yaw, design an adaptive weight controller based on fuzzy logic and generate the optimal weight matrix.
[0009] Step S4: Based on the lateral and longitudinal coupled vehicle state space equation considering yaw and the optimal weight matrix, design a lateral and longitudinal coupled model predictive controller to generate the target front wheel steering angle and the target longitudinal acceleration.
[0010] Step S5: Based on the target front wheel angle and target longitudinal acceleration, a lateral and longitudinal collaborative decision-making and torque distribution layer is constructed to integrate and coordinate the target front wheel angle and target longitudinal acceleration generated in step S4, thereby generating the final actuator control signal to realize the control of the distributed drive intelligent vehicle.
[0011] The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction provided by the present invention has the following beneficial effects:
[0012] (1) By establishing a vehicle state space equation that considers yaw, this invention solves the problem of decreased trajectory tracking accuracy caused by neglecting the mutual influence of yaw and longitudinal motion in complex dynamic scenarios under traditional decoupling control, and significantly improves the path following capability under extreme conditions.
[0013] (2) The present invention introduces an adaptive weight controller based on fuzzy logic, which realizes the online dynamic adjustment of the multi-objective optimization weight of model predictive control (MPC), enabling the controller to adapt to changes in vehicle parameters and diverse driving conditions in real time, greatly enhancing the robustness and generalization ability of the system, thereby improving vehicle control performance.
[0014] (3) This invention utilizes the rolling optimization and constraint processing capabilities of the coupled model predictive controller to simultaneously optimize the front wheel angle and longitudinal acceleration in the prediction time domain, generating the optimal control sequence that satisfies the dynamic constraints, effectively ensuring the driving stability of the vehicle in high-speed scenarios. Attached Figure Description
[0015] Figure 1 This is a flowchart illustrating the distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction provided by the present invention.
[0016] Figure 2 This is a schematic diagram comparing the position errors of three control methods in a real-vehicle test on a double lane change road.
[0017] Figure 3 This is a schematic diagram comparing the lateral errors of three control methods in a real-vehicle test on a double lane change road.
[0018] Figure 4 This is a schematic diagram comparing the heading errors of three control methods in a real-vehicle test on a double lane change road.
[0019] Figure 5 This is a schematic diagram comparing the speed tracking errors of three control methods in a real-vehicle test on a double lane change road.
[0020] Figure 6 This is a schematic diagram comparing the center of gravity sideslip angle of three control methods in a real-vehicle test on a double lane change road. Detailed Implementation
[0021] To facilitate understanding of the present invention, a more complete description will be given below with reference to various embodiments. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.
[0023] Please see Figure 1 The embodiments of the present invention provide a distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction, including steps S1-S5:
[0024] Step S1: Acquire multi-source vehicle data of the distributed drive intelligent vehicle through on-board sensors. The multi-source vehicle data includes vehicle dynamics data and road environment data.
[0025] Specifically, vehicle dynamics data includes: vehicle mass The yaw moment of inertia of a vehicle about an axis passing through its center of mass and perpendicular to the ground. Speed Longitudinal speed Lateral speed yaw rate lateral angle lateral stiffness of the front wheel lateral stiffness of the rear wheel Front wheel steering angle Actual acceleration The horizontal and vertical coordinates of the vehicle's center of mass Distance from center of gravity to front axle Distance from center of gravity to rear axle Wheelbase Center of mass height Front axle track and rear axle track .
[0026] Road environment data includes: aiming points Reference trajectory at the pre-aiming point curvature at Reference trajectory at the pre-aiming point The heading angle of the tangent Heading angle rate of change .
[0027] Step S2: Based on multi-source vehicle data, establish a vehicle state-space equation that considers yaw and is coupled lateral and longitudinal directions.
[0028] Specifically, step S2 includes:
[0029] First, calculate the heading error respectively. Lateral displacement error Horizontal displacement error change rate Yaw rate error Speed error The expression is:
[0030]
[0031]
[0032]
[0033]
[0034]
[0035] in, Pre-aiming point The x and y coordinates, For the desired vehicle speed;
[0036] Then, the lateral and longitudinal coupled vehicle state-space equations considering yaw are established, expressed as:
[0037]
[0038]
[0039]
[0040]
[0041] in, Let be the state variables in the vehicle's state-space equation. for The first-order differential, These are the input variables for the vehicle state-space equations. The output variable of the vehicle state-space equation is: , Indicates transpose. This represents the desired longitudinal acceleration; , , , The intermediate matrices for the horizontally and vertically coupled vehicle state-space equations are as follows:
[0042]
[0043]
[0044]
[0045]
[0046] in, The inertial time constant, This represents the system gain.
[0047] Step S3: Based on the lateral and longitudinal coupled vehicle state-space equations considering yaw, design an adaptive weight controller based on fuzzy logic to generate the optimal weight matrix.
[0048] Specifically, step S3 includes:
[0049] Based on the lateral displacement error and its rate of change, the heading error and yaw rate error, the velocity error and its rate of change, and the road curvature, fuzzy controllers one, two, three, and four are designed respectively. The domain range of the input variables for each fuzzy controller is set as follows:
[0050] For fuzzy controller one, , For input, For output; set the lateral displacement error. Domain of discourse is lateral displacement error change rate Domain of discourse is ,in, , , , , These are fuzzy linguistic variables, representing negative large, negative small, zero, positive small, and positive large, respectively.
[0051] For fuzzy controller two, , For input, For output; set the heading error The domain is Yaw rate error The domain is ;
[0052] For fuzzy controller three, , As input, with For output, It is the rate of change of speed error; set the speed error. The domain is Rate of change of speed error The domain is ;
[0053] For fuzzy controller four, As input, with , , For output; set curvature The domain is ;
[0054] Set the range of each output universe to:
[0055]
[0056]
[0057]
[0058]
[0059]
[0060]
[0061] in, , These are fuzzy linguistic variables, representing positive minimum and positive middle, respectively;
[0062] For the input of the fuzzy controller The following equation is satisfied:
[0063]
[0064] in, For Gaussian membership functions, or or As input to the fuzzy controller , , , for The first-order differential, Let be the standard deviation of the function. The central value of the function;
[0065] For the output of the fuzzy controller The following equation is satisfied:
[0066]
[0067] in, Let be the membership function of the triangle, or or or or or As the output of the fuzzy controller ; This is the starting point on the left side of the function, and the point whose membership degree starts to increase from 0. The vertex of the function is the peak point with a membership degree of 1; The point where the membership degree drops to 0 is the endpoint on the right side of the function;
[0068] Then design the input of the fuzzy controller. , With output , , The fuzzy rules between them, and the input of the fuzzy controller. With output , , The fuzzy rules between them;
[0069] The fuzzy rule tables for error and weight coefficients are shown in Tables 1 to 3. The fuzzy rule table for road curvature and weight ratio coefficients is shown in Table 4.
[0070] Table 1
[0071]
[0072] Table 2
[0073]
[0074] Table 3
[0075]
[0076] Table 4
[0077]
[0078] Then the weight matrix Diagonal coefficient The optimal control law is designed as follows:
[0079]
[0080]
[0081]
[0082] in, , , The physical meaning is that it represents the weight coefficient value after being processed by the fuzzy controller one, two, and three. , , The physical meaning of is the proportional coefficient of road curvature.
[0083] Finally, the optimal weight matrix is obtained. for:
[0084] .
[0085] Step S4: Based on the lateral and longitudinal coupled vehicle state space equations considering yaw and the optimal weight matrix, design a lateral and longitudinal coupled model predictive controller to generate the target front wheel steering angle and the target longitudinal acceleration.
[0086] Specifically, step S4 includes:
[0087] First, a zero-order hold method is used at each time step to discretize the state-space equations of the laterally and longitudinally coupled vehicle, resulting in the following discrete vehicle state-space equations:
[0088]
[0089]
[0090] in, for Time-state variables The value, for Time-state variables The value, for Input variables at all times The value, for Output variables at all times The value, The state matrix, , It is the identity matrix. The sampling time interval, For the control matrix, , The steady-state error matrix is... , For the output matrix, ;
[0091] Then, an error state variable is introduced, leading to the augmented discrete vehicle state-space equation, expressed as:
[0092]
[0093]
[0094] in, express Time error state variables The value, express Time error state variables The value, , for Input variables at all times The value, for time The value, for Input variables at all times The increment, , and An intermediate matrix for augmenting the discrete vehicle state-space equations;
[0095]
[0096]
[0097]
[0098] Then, an objective function is established that includes two evaluation indicators: control increment and tracking error. The expression is:
[0099]
[0100] in, To predict the window point, To control the window point, and , For the future The output variable predicted by the time step, For the future The control increment predicted by the time step, It is a diagonal positive definite weight matrix. As a relaxation factor, These are the weighting coefficients of the relaxation factor;
[0101]
[0102] in, , , Each of the future The time step predicted , , ;
[0103]
[0104] in, , Each of the future The predicted front wheel steering angle increment and the expected longitudinal acceleration increment at the time step and The weights correspond to the front wheel steering angle and the desired acceleration, respectively.
[0105] Redefine the total input increment vector for the current and future time steps. State vector and output vector , respectively represented as:
[0106]
[0107]
[0108]
[0109] in, for Input variables at all times The increment, for Input variables at all times The increment, for Time error state variables The value, for Output variables at all times The value, for Output variables at all times The value;
[0110] Then, the state in the prediction time domain is calculated iteratively based on the augmented discrete vehicle state-space equation, and the expression is:
[0111]
[0112] in, for Time error state variables The value;
[0113] Based on the iteration, the predicted output in the time domain is obtained as follows:
[0114]
[0115] in, for Output variables at all times The value;
[0116] The future prediction output will be:
[0117]
[0118] in, and This is the process matrix;
[0119]
[0120]
[0121] Then the objective function is updated to obtain the updated objective function. The expression is:
[0122]
[0123] in, The total state vector containing coefficients. This is the total quadratic coefficient matrix. This is the vector of coefficients for the first-order terms;
[0124]
[0125]
[0126]
[0127] Finally, by minimizing under the set constraints... To obtain the optimal control increment This allows us to obtain the target front wheel steering angle. longitudinal acceleration relative to the target .
[0128] It should be noted that in the lateral tracking test of the vehicle, there are left and right extreme positions for the front wheel turning angle or steering wheel turning angle; in order to ensure that the vehicle does not experience a large sense of jerking during the longitudinal movement of the vehicle, it is necessary to consider the acceleration constraint; in addition, it is also necessary to ensure that the constraint conditions are met in all time steps.
[0129] Specifically, the constraints include:
[0130]
[0131]
[0132]
[0133]
[0134]
[0135]
[0136] in, and Input variables The minimum and maximum values, and These are the minimum and maximum values of the front wheel steering angle, respectively. and These are the minimum and maximum values of the desired longitudinal acceleration, respectively. To control the input, for Input variables at all times The value, for Input variables at all times The value, and They are respectively The minimum and maximum values, and These are the minimum and maximum values of the total input increment vector, respectively. and These are the minimum and maximum values of the output vector, respectively.
[0137] Step S5: Based on the target front wheel angle and target longitudinal acceleration, a lateral and longitudinal collaborative decision-making and torque distribution layer is constructed to integrate and coordinate the target front wheel angle and target longitudinal acceleration generated in step S4, thereby generating the final actuator control signal to realize the control of the distributed drive intelligent vehicle.
[0138] Specifically, the target front wheel steering angle calculated in step S4 is... The signal is sent directly to the steering servo controller to drive the front wheels to steer, thus increasing the longitudinal acceleration of the target. Substituting the values into the vehicle's inverse longitudinal dynamics model, the total driving torque is obtained. The expression for the vehicle's inverse longitudinal dynamics model is:
[0139]
[0140] in, For driving resistance, The radius is the wheel radius.
[0141] Then the total driving torque The system is divided into four equal parts, which are then sent to four drive motor controllers. The vehicle actuators respond accordingly, thus achieving distributed drive intelligent vehicle control.
[0142] Verification Case: In a real-vehicle test scenario with double lane change, the control effects of the three methods were compared. The method of this invention is named "CP MPC", the hierarchical architecture MPC coupling control method in the prior art is named "LY MPC", and the hierarchical parallel LQR-PID coupling control method in the prior art is named "LY LQR PID".
[0143] The autonomous vehicle is set to enter the double lane change trajectory with an initial speed of 5 km / h, and then accelerate to 54 km / h and maintain a constant speed.
[0144] like Figure 2 As shown, all three methods can basically track the reference trajectory, but there are differences in lateral error, heading error, and centroid sideslip. Figure 3 , Figure 4 , Figure 6 As shown, during lane change, the LY LQR PID controller, in order to ensure stable heading and sideslip angle, exhibited a lateral displacement error exceeding 0.1m, and upon exiting the curve, the lateral displacement error reached 0.14m, resulting in significant yaw. It took 3 seconds to correct back to the reference path. While the LY MPC controller could maintain a stable lateral error within 0.1m, its layered control strategy for both lateral and longitudinal directions led to significant fluctuations in lateral error as vehicle speed changed. The lateral and longitudinal controllers could not guarantee optimal control simultaneously. Furthermore, the LY MPC controller exhibited a large heading error exceeding 0.1rad and also showed slight sideslip. After completing the lane change, the vehicle took 3-4 seconds to return to the reference trajectory, which may affect driving comfort. In this case, the CP MPC proposed in this invention exhibits excellent coupled control performance and strong robustness. During trajectory tracking, the motion trajectory basically follows the reference trajectory, with a maximum lateral displacement error of only 0.0391m and a heading error within 0.1rad. Furthermore, after completing the lane change maneuver, due to the adaptive parameter adjustment of fuzzy control, the vehicle can quickly maintain lateral and heading stability and achieve zero steady-state error.
[0145] In addition, such as Figure 5 As shown, the method of this invention and LY MPC can stably track the reference speed and ensure a small overshoot while keeping the steady-state error within 0.1m / s when the speed changes rapidly, demonstrating the excellent performance of MPC. In contrast, LY LQR PID, due to the use of PID control in longitudinal control, exhibits obvious jitter in speed tracking after reaching the target vehicle speed due to lateral disturbances.
[0146] In summary, the distributed drive intelligent vehicle control method based on the lateral and longitudinal coupling model prediction described above has the following beneficial effects:
[0147] (1) By establishing a vehicle state space equation that considers yaw, this invention solves the problem of decreased trajectory tracking accuracy caused by neglecting the mutual influence of yaw and longitudinal motion in complex dynamic scenarios under traditional decoupling control, and significantly improves the path following capability under extreme conditions.
[0148] (2) The present invention introduces an adaptive weight controller based on fuzzy logic, which realizes the online dynamic adjustment of the multi-objective optimization weight of model predictive control (MPC), enabling the controller to adapt to changes in vehicle parameters and diverse driving conditions in real time, greatly enhancing the robustness and generalization ability of the system, thereby improving vehicle control performance.
[0149] (3) This invention utilizes the rolling optimization and constraint processing capabilities of the coupled model predictive controller to simultaneously optimize the front wheel angle and longitudinal acceleration in the prediction time domain, generating the optimal control sequence that satisfies the dynamic constraints, effectively ensuring the driving stability of the vehicle in high-speed scenarios.
Claims
1. A distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction, characterized in that, include: Step S1: Acquire multi-source vehicle data of the distributed drive intelligent vehicle through on-board sensors. The multi-source vehicle data includes vehicle dynamics data and road environment data. Step S2: Based on multi-source vehicle data, establish a lateral and longitudinal coupled vehicle state-space equation that considers yaw. Step S3: Based on the lateral and longitudinal coupled vehicle state space equation considering yaw, design an adaptive weight controller based on fuzzy logic and generate the optimal weight matrix. Step S4: Based on the lateral and longitudinal coupled vehicle state space equation considering yaw and the optimal weight matrix, design a lateral and longitudinal coupled model predictive controller to generate the target front wheel steering angle and the target longitudinal acceleration. Step S5: Based on the target front wheel angle and target longitudinal acceleration, a lateral and longitudinal collaborative decision-making and torque distribution layer is constructed to integrate and coordinate the target front wheel angle and target longitudinal acceleration generated in step S4, thereby generating the final actuator control signal to realize the control of the distributed drive intelligent vehicle.
2. The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction according to claim 1, characterized in that, In step S1, the vehicle dynamics data includes: vehicle mass The yaw moment of inertia of a vehicle about an axis passing through its center of mass and perpendicular to the ground. Speed Longitudinal speed Lateral speed yaw rate lateral angle lateral stiffness of the front wheel lateral stiffness of the rear wheel Front wheel steering angle Actual acceleration The horizontal and vertical coordinates of the vehicle's center of mass Distance from center of gravity to front axle Distance from center of gravity to rear axle Wheelbase Center of mass height Front axle track and rear axle track ; Road environment data includes: aiming points Reference trajectory at the pre-aiming point curvature at Reference trajectory at the pre-aiming point The heading angle of the tangent Heading angle rate of change .
3. The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction according to claim 2, characterized in that, Step S2 specifically includes: First, calculate the heading error respectively. Lateral displacement error Horizontal displacement error change rate Yaw rate error Speed error The expression is: in, Pre-aiming point The x and y coordinates, For the desired vehicle speed; Then, the lateral and longitudinal coupled vehicle state-space equations considering yaw are established, expressed as: in, Let be the state variables in the vehicle's state-space equation. for The first-order differential, These are the input variables for the vehicle's state-space equations. For the output variables of the vehicle state-space equations, Indicates transpose. This represents the desired longitudinal acceleration; , , , The intermediate matrices for the horizontally and vertically coupled vehicle state-space equations are as follows: in, The inertial time constant, This represents the system gain.
4. The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction according to claim 3, characterized in that, Step S3 specifically includes: Based on the lateral displacement error and its rate of change, the heading error and yaw rate error, the velocity error and its rate of change, and the road curvature, fuzzy controllers one, two, three, and four are designed respectively. The domain range of the input variables for each fuzzy controller is set as follows: For fuzzy controller one, , For input, For output; set the lateral displacement error. Domain of discourse is lateral displacement error change rate Domain of discourse is ,in, , , , , These are fuzzy linguistic variables, representing negative large, negative small, zero, positive small, and positive large, respectively. For fuzzy controller two, , For input, For output; set the heading error Domain of discourse is Yaw rate error Domain of discourse is ; For fuzzy controller three, , As input, with For output, It is the rate of change of speed error; set the speed error. Domain of discourse is Rate of change of speed error Domain of discourse is ; For fuzzy controller four, As input, with , , For output; set curvature Domain of discourse is ; Set the range of each output universe of discourse as follows: in, , These are fuzzy linguistic variables, representing positive minimum and positive middle, respectively; For the input of the fuzzy controller The following equation is satisfied: in, For Gaussian membership functions, or or As input to the fuzzy controller , , , for The first-order differential, Let be the standard deviation of the function. The central value of the function; For the output of the fuzzy controller The following equation is satisfied: in, Let be the membership function of the triangle, or or or or or As the output of the fuzzy controller ; The starting point on the left side of the function; The vertex of the function; The endpoint on the right side of the function; Then design the input of the fuzzy controller. , With output , , The fuzzy rules between them, and the input of the fuzzy controller. With output , , The fuzzy rules between them; Then the weight matrix Diagonal coefficient The optimal control law is designed as follows: Finally, the optimal weight matrix is obtained. for: 。 5. The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction according to claim 4, characterized in that, Step S4 specifically includes: First, a zero-order hold method is used at each time step to discretize the state-space equations of the laterally and longitudinally coupled vehicle, resulting in the following discrete vehicle state-space equations: in, for Time-state variables The value, for Time-state variables The value, for Input variables at all times The value, for Output variables at all times The value, The state matrix, , It is the identity matrix. The sampling time interval, For the control matrix, , The steady-state error matrix is... , For the output matrix, ; Then, an error state variable is introduced, leading to the augmented discrete vehicle state-space equation, expressed as: in, express Time error state variables The value, express Time error state variables The value, , for Input variables at all times The value, for time The value, for Input variables at all times The increment, , and An intermediate matrix for augmenting the discrete vehicle state-space equations; Then, an objective function is established that includes two evaluation indicators: control increment and tracking error. The expression is: in, To predict the window point, To control the window point, and , For the future The output variable predicted by the time step, For the future The control increment predicted by the time step, It is a diagonal positive definite weight matrix. As a relaxation factor, These are the weighting coefficients of the relaxation factor; in, , , Each of the future The time step predicted , , ; in, , Each of the future The predicted front wheel steering angle increment and the expected longitudinal acceleration increment at the time step and The weights correspond to the front wheel steering angle and the desired acceleration, respectively. Redefine the total input increment vector for the current and future time steps. State vector and output vector , respectively represented as: in, for Input variables at all times The increment, for Input variables at all times The increment, for Time error state variables The value, for Output variables at all times The value, for Output variables at all times The value; Then, the state in the prediction time domain is calculated iteratively based on the augmented discrete vehicle state-space equation, and the expression is: in, for Time error state variables The value; Based on the iteration, the predicted output in the time domain is obtained as follows: in, for Output variables at all times The value; The future prediction output will be: in, and This is the process matrix; Then the objective function is updated to obtain the updated objective function. The expression is: in, The total state vector containing coefficients. This is the total quadratic coefficient matrix. This is the vector of coefficients for the first-order terms; Finally, by minimizing under the set constraints... To obtain the optimal control increment This allows us to obtain the target front wheel steering angle. longitudinal acceleration relative to the target .
6. The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction according to claim 5, characterized in that, The constraints include: in, and Input variables The minimum and maximum values, and These are the minimum and maximum values of the front wheel steering angle, respectively. and These are the minimum and maximum values of the desired longitudinal acceleration, respectively. To control the input, for Input variables at all times The value, for Input variables at all times The value, and They are respectively The minimum and maximum values, and These are the minimum and maximum values of the total input increment vector, respectively. and These are the minimum and maximum values of the output vector, respectively.
7. The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction according to claim 5, characterized in that, Step S5 specifically includes: The target front wheel steering angle calculated in step S4 The signal is sent directly to the steering servo controller to drive the front wheels to steer, thus increasing the longitudinal acceleration of the target. Substituting the values into the vehicle's inverse longitudinal dynamics model, the total driving torque is obtained. Then the total driving torque The system is divided into four equal parts, which are then sent to four drive motor controllers. The vehicle actuators respond accordingly, thus achieving distributed drive intelligent vehicle control.
8. The distributed drive intelligent vehicle control method based on lateral and longitudinal coupling model prediction according to claim 7, characterized in that, The expression for the vehicle's inverse longitudinal dynamics model is: in, For driving resistance, The radius is the wheel radius.