A rear wheel steering and active stabilizer bar collaborative control method and system

Abstract

The application discloses a rear wheel steering and active stabilizer bar cooperative control method, which comprises the following steps: collecting vehicle state signals and driver instructions; using an extended Kalman filter to estimate the state vector of the current vehicle; judging the current working condition of the vehicle through multiple conditions based on the vehicle state signals, the driver instructions and a preset rule base; using a linear parameter time-varying model as a prediction model of the vehicle state to solve an optimal control sequence through model predictive control; in the model predictive control, the attitude deviation cost and the actuator action cost of the objective function are adjusted based on the current working condition of the vehicle, and the matrix parameters of the linear parameter time-varying model are adjusted based on the vehicle speed; and the optimal control instruction obtained based on the optimal control sequence is executed by an active rear wheel steering system and an active lateral stabilizer system. The application further discloses a corresponding control system. The application realizes the cooperative decision of ARS and AAS, and can realize high-precision, high-smoothness and high-stability vehicle attitude control in all working conditions.

Description

Technical Field

[0001] This invention relates to the field of control system technology, and in particular to a method and system for coordinated control of rear wheel steering and active stabilizer bar. Background Technology

[0002] With the development of intelligent and electric vehicles, advanced driver assistance systems (ADAS) and users' demands for the ultimate driving experience have placed higher requirements on vehicle chassis control. The handling stability and ride comfort of a vehicle largely depend on the chassis system's ability to control the vehicle's posture.

[0003] The vehicle utilizes chassis actuators such as Active Rear-Wheel Steering (ARS) and Active Anti-Rotation Bar (AAS). The ARS system achieves independent control of the rear wheel steering angle by adding a steering actuator to the rear axle. Typically, at low speeds, the rear wheels rotate in the opposite direction to the front wheels to reduce the turning radius, while at high speeds they rotate in the same direction to improve lane-changing stability. The AAS system, on the other hand, replaces traditional torsion bar springs with electric motors or hydraulic mechanisms, actively generating anti-roll torque to significantly suppress body roll during cornering.

[0004] However, existing technologies have the following inherent drawbacks:

[0005] 1) Control island problem: ARS and AAS are controlled by different electronic control units (ECUs), each with its own independent control objectives (ARS mainly focuses on yaw rate and trajectory tracking, while AAS mainly focuses on roll angle). This "each fighting its own battle" control strategy cannot achieve optimal global performance and may even lead to control conflicts under extreme conditions.

[0006] 2) Response lag: Existing control is mostly based on post-event control based on feedback from vehicle status sensors, which lacks the ability to predict the driver's intentions and the future vehicle status, resulting in a delay in control response.

[0007] 3) Functional limitation: The existing control strategies of ARS and AAS are mainly designed for a single cornering condition, and fail to fully utilize the synergistic potential of the two actuators in suppressing pitch (acceleration / braking head nod), vertical sway, and resisting external disturbances (such as crosswinds).

[0008] 4) Performance compromise: Over-reliance on AAS in pursuit of anti-roll performance will lead to a stiffer suspension equivalent stiffness, sacrificing comfort when driving straight and crossing single-sided obstacles. Summary of the Invention

[0009] To address the shortcomings of the existing technology, this invention provides a rear-wheel steering and active stabilizer bar coordinated control method, achieving coordinated decision-making between ARS and AAS, controlling comprehensive stability in the lateral, longitudinal, and vertical directions, and maximizing vehicle handling limits and stability, as well as improving system response speed and smoothness without excessively sacrificing comfort. This invention also provides a rear-wheel steering and active stabilizer bar coordinated control system.

[0010] The technical solution of the present invention is as follows:

[0011] A method for coordinated control of rear-wheel steering and active stabilizer bar includes:

[0012] Collect vehicle status signals and driver commands;

[0013] Extended Kalman filtering is used to estimate the current vehicle state vector, which includes roll angle, roll rate, pitch angle, and pitch rate, based on the vehicle state signal.

[0014] The current operating condition of the vehicle is determined by multiple conditions based on vehicle status signals, driver commands, and a pre-set rule base.

[0015] Model predictive control is performed using a linear parameter time-varying model as the predictive model for vehicle state: Based on the current vehicle state vector and driver commands, the objective function is solved under the system dynamics constraints, actuator physical constraints, and rate constraints of the linear parameter time-varying model to obtain the optimal control sequence for the rear wheel steering angle, front axle active stabilizer bar torque, and rear axle active stabilizer bar torque. The objective function is to minimize the sum of attitude deviation cost, actuator action cost, and terminal state deviation cost in the finite time domain. The matrix parameters of the linear parameter time-varying model are adjusted based on vehicle speed, and the attitude deviation cost and actuator action cost are adjusted based on the current operating conditions of the vehicle.

[0016] The optimal control command obtained based on the optimal control sequence is executed by the active rear wheel steering system and the active lateral stabilizer bar system to perform rear wheel steering angle tracking and output the target anti-roll moment.

[0017] Furthermore, the vehicle's operating modes include high-speed cornering / emergency lane change, emergency braking, bumpy road surface, and strong crosswind / single-sided adhesion. The attitude deviation cost and actuator action cost are adjusted based on the vehicle's current operating conditions by setting calculation weight benchmark values ​​for attitude deviation cost and actuator action cost for each operating condition. Then, based on the continuous confidence of the operating mode to which the vehicle's current operating condition belongs, output by the fuzzy logic algorithm, the calculation weight benchmark values ​​for attitude deviation cost and actuator action cost are smoothly interpolated to obtain the calculation weights for attitude deviation cost and actuator action cost.

[0018] Furthermore, the objective function is expressed as follows:

[0019] ,

[0020] Where U is the control sequence to be optimized. Representing a finite time domain, , The vehicle's roll angle. The vehicle's pitch angle. , The weights for calculating the attitude deviation cost are... For control vectors, The weights for calculating the cost of actuator actions. Let be the vehicle's state vector. , The weights for calculating the terminal state deviation cost.

[0021] Furthermore, the calculation weight benchmark value for the attitude deviation cost under the high-speed cornering / emergency lane change condition is... The baseline value for calculating the actuator action cost is... ([50, 10, 10]); The baseline weight for calculating the attitude deviation cost under emergency braking conditions is [50, 10, 10]. The baseline value for calculating the actuator action cost is... ([20, 50, 50]); The benchmark weight for calculating the attitude deviation cost under bumpy road conditions is [value missing]. The baseline value for calculating the actuator action cost is... ([10, 500, 500]); The baseline weight for calculating the attitude deviation cost under strong crosswind / single-sided adhesion conditions is [value missing]. The baseline value for calculating the actuator action cost is... ([30, 50, 50]).

[0022] Furthermore, the linear parameter time-varying model is expressed as follows:

[0023] ,

[0024] in, This represents the vehicle's state vector. For control vectors, For driver instructions, , and For matrix parameters, Indicates the time.

[0025] Furthermore, the matrix parameters of the linear parameter time-varying model are adjusted based on vehicle speed by constructing a vehicle speed-matrix parameter lookup table offline, and then calculating the matrix parameters at the current vehicle speed through linear interpolation based on the matrix parameter values ​​corresponding to a specific vehicle speed in the lookup table.

[0026] Furthermore, the offline construction of the vehicle speed-matrix parameter lookup table includes the following steps:

[0027] A linear parameter time-varying model matching the target vehicle model is built in the simulation software, and a sweep frequency signal is applied as an excitation to the driver's command;

[0028] Noise filtering is used to remove measurement noise of the vehicle state obtained from a more dynamic model, and the vehicle state and driver commands are normalized.

[0029] The linear parameter time-varying model is discretized into an identification standard form, and then the matrix parameters of the corresponding linear parameter time-varying model at the current vehicle speed are obtained by using the recursive least squares method until the parameters converge.

[0030] Another technical solution of the present invention is: a rear-wheel steering and active stabilizer bar coordinated control system, comprising:

[0031] The data acquisition module is used to collect vehicle status signals and driver commands;

[0032] The state estimation module is used to estimate the current vehicle state vector, including roll angle, roll rate, pitch angle, and pitch rate, based on the vehicle state signal using extended Kalman filtering.

[0033] The operating condition estimation module is used to determine the current operating condition of the vehicle based on vehicle status signals, driver commands, and a pre-set rule base through multiple conditions.

[0034] The predictive control module is used to perform model predictive control based on a linear parametric time-varying model as the predictive model for vehicle state. Based on the current vehicle state vector, it solves for the optimal control sequence of rear wheel steering angle, front axle active stabilizer bar torque, and rear axle active stabilizer bar torque by solving the objective function under system dynamics constraints, actuator physical constraints, and rate constraints of the linear parametric time-varying model. The objective function minimizes the sum of attitude deviation cost, actuator action cost, and terminal state deviation cost within a finite time domain. The matrix parameters of the linear parametric time-varying model are adjusted based on vehicle speed, and the attitude deviation cost and actuator action cost are adjusted based on the vehicle's current operating conditions.

[0035] The execution module includes an active rear-wheel steering system and an active lateral stabilizer bar system, which are used to execute the optimal control command obtained based on the optimal control sequence, perform rear wheel steering angle tracking, and output the target anti-roll moment.

[0036] Compared with the prior art, the present invention has the following advantages:

[0037] By breaking down the control silos between the active rear-wheel steering (ARS) and the active lateral stabilizer bar (AAS), a deep collaborative control framework for the two was established. Integrated collaborative decision-making was achieved through unified MPC control, thus integrating the performance of the subsystems.

[0038] By expanding the control objective from a single roll stability to a two-dimensional vehicle angle attitude stability encompassing both roll and pitch, and by using a linear time-varying model to adapt to vehicle speed and cost weights to adapt to operating conditions, and by collaboratively optimizing attitude tracking, actuator smoothness, and terminal state convergence within a finite time domain, high-precision, high-smoothness, and high-stability vehicle attitude control can be achieved under all operating conditions, providing a good overall driving experience. Attached Figure Description

[0039] Figure 1 This is a schematic diagram of the hardware architecture of the rear wheel steering and active stabilizer bar coordinated control system as an example.

[0040] Figure 2 This is a flowchart illustrating the rear-wheel steering and active stabilizer bar coordinated control method in an embodiment.

[0041] Figure 3 The figure shows the simulation results of the rear wheel steering and active stabilizer bar coordinated control method of the present invention and the traditional independent control method. Detailed Implementation

[0042] The present invention will be further described below with reference to embodiments, but these are not intended to limit the scope of the invention.

[0043] The rear-wheel steering and active stabilizer bar coordinated control system of this embodiment mainly includes a data acquisition module, a state estimation module, a working condition estimation module, a predictive control module, and an execution module.

[0044] The data acquisition module is used to collect vehicle status signals and driver commands.

[0045] The state estimation module is used to estimate the current vehicle state vector, including roll angle, roll rate, pitch angle, and pitch rate, based on the vehicle state signal using extended Kalman filtering.

[0046] The operating condition estimation module is used to determine the current operating condition of the vehicle based on vehicle status signals, driver commands, and a pre-set rule base through multiple conditions.

[0047] The predictive control module is used to perform model predictive control based on a linear parameter time-varying model as the predictive model of the vehicle state. Based on the current vehicle state vector, the objective function is solved under the system dynamics constraints, actuator physical constraints, and rate constraints of the linear parameter time-varying model to obtain the optimal control sequence of the rear wheel steering angle, front axle active stabilizer bar torque, and rear axle active stabilizer bar torque. The objective function is to minimize the sum of attitude deviation cost, actuator action cost, and terminal state deviation cost in the finite time domain. The matrix parameters of the linear parameter time-varying model are adjusted based on the vehicle speed, and the attitude deviation cost and actuator action cost are adjusted based on the current operating conditions of the vehicle.

[0048] The execution module, including the active rear-wheel steering system and the active lateral stabilizer bar system, is used to execute the optimal control command based on the optimal control sequence, perform rear wheel steering angle tracking, and output the target anti-roll moment.

[0049] The hardware architecture of the system is as follows Figure 1 As shown, it includes:

[0050] The sensing module is used to collect vehicle status signals and driver commands in real time. It includes:

[0051] Steering wheel angle sensor, vehicle speed sensor, inertial measurement unit (IMU, used to measure vehicle yaw rate, lateral acceleration, longitudinal acceleration, roll angle, pitch angle and their respective angular velocities), and four wheel speed sensors.

[0052] The execution module includes:

[0053] 1) Active Rear Wheel Steering System (ARS): Includes the rear wheel steering ECU, steering actuator motor, and steering transmission mechanism. It receives the target rear wheel steering angle command. And drive the rear wheels to achieve precise cornering tracking;

[0054] 2) Active Stabilizer System (AAS): This includes the front / rear axle active stabilizer bar ECU, torque actuator (such as a motor), and the front and rear stabilizer bar bodies. It receives the target stabilizer bar torque command. (former) and (After) and output precise torque.

[0055] Core processing module – Cooperative Controller:

[0056] This controller is part of the chassis domain controller (VDC) or the vehicle's central computing platform and features a high-performance microprocessor. It communicates with the sensing and actuation modules via a vehicle bus (such as CAN FD or Automotive Ethernet) and implements a method for coordinated control of rear-wheel steering and active stabilizer bars.

[0057] The collaborative controller is configured to perform the following core functions:

[0058] 1) Receive all signals from the sensing module;

[0059] 2) Based on the received signals, the real-time state vector of the current vehicle is estimated using the vehicle state observer. Including: centroid sideslip angle Tire vertical load (i=f,r; j=l,r), estimated value of road surface adhesion coefficient μ;

[0060] 3) Based on driver commands (steering wheel angle) Accelerator pedal opening A cc Brake pedal opening B rk The system uses the vehicle's status and condition to determine the vehicle's current overall operating conditions (e.g., steady-state cornering, emergency lane change, emergency braking, full acceleration, bumpy road surface, strong crosswind, etc.).

[0061] 4) Based on the comprehensive operating conditions, invoke the cooperative control algorithm to solve for the optimal active rear wheel steering angle in an integrated manner. With active stabilizer bar torque (forward) , (back);

[0062] 5) Calculate the optimal instruction. , , Send to ARS and AAS actuators.

[0063] For the coordinated control method implemented by the rear-wheel steering and active stabilizer bar coordinated control system, please refer to [link to relevant documentation]. Figure 2 As shown, it includes:

[0064] 1) System initialization:

[0065] After power-on, load the vehicle's inherent parameters (mass, moment of inertia, wheelbase, tire lateral stiffness, etc.); load the initial values ​​of the MPC weight matrices Q and R, actuator constraint limits, and the operating condition identification rule base; pre-calculate the MPC terminal weight matrix P (by solving the Riccati equation).

[0066] 2) Real-time control loop (period: 10ms):

[0067] a. Data acquisition and preprocessing:

[0068] All sensor data, including vehicle status signals and driver commands, is periodically received via the CAN FD bus, and low-pass filtered and validated. If any sensor data is abnormal, an estimated value or a default safety value is used.

[0069] b. Vehicle status monitoring:

[0070] High-precision state estimation is achieved using the Extended Kalman Filter (EKF). The EKF employs a more complete vehicle dynamics model (including yaw motion) than the MPC prediction model, and its state vector can contain... The observables are those that can be directly measured. , , EKF, by fusing observation information, outputs the parameters required for MPC. The estimated value can also provide the centroid sideslip angle. Auxiliary information, etc. For the vehicle's roll angle, For the vehicle's roll rate, For the vehicle's pitch angle, Let ω be the vehicle's pitch angular velocity. Among them, roll angle and pitch angle are the final attitude quantities that need to be stabilized directly, while their angular velocities serve as intermediate states to ensure the smoothness of control and dynamic performance.

[0071] c. Comprehensive operating condition identification:

[0072] The vehicle's current operating condition is determined based on multiple conditions, including vehicle status signals, driver commands, a pre-set rule base, and fuzzy logic.

[0073] High-speed cornering / emergency lane change conditions: If the steering wheel angular velocity lateral acceleration And the brake pedal opening is less than 10%;

[0074] Emergency braking condition: If the master cylinder pressure > 80 bar or the brake pedal opening > 90%, and > 30km / h;

[0075] Bumpy road conditions: If the high-frequency (5-15Hz) vibration energy of the four wheel speed sensors is significantly higher than the baseline;

[0076] Strong crosswind / single-sided adhesion conditions: If the steering wheel angular velocity is <10deg / s and the absolute steering wheel angle is <5°, the lateral acceleration is >0.15g, the vehicle speed is ≥60km / h, and the brake / accelerator pedal opening is <20%.

[0077] A fuzzy logic algorithm is used to output the confidence level of each working condition, which is then used to smoothly adjust the MPC weights.

[0078] d. Using a time-varying linear parameter model as the predictive model for vehicle state, perform model predictive control:

[0079] First, a predictive model is constructed, employing a simplified linear parametric time-varying (LPV) model oriented towards vehicle attitude control. This model equivalently characterizes the dynamic influence of actuator control quantity u and driver disturbance input w on the core vehicle attitude state x.

[0080]

[0081] Where: state vector Control vector This refers to the rear wheel steering angle and the torques of the front and rear stabilizer bars. The disturbance vector can be measured. This represents the primary driver inputs: front wheel steering angle and vehicle longitudinal acceleration. Front wheel steering angle From the steering wheel angle via steering ratio The conversion yields: In this embodiment =16 (can be adapted to different vehicle models). This enables the MPC to provide feedforward compensation for the driver's steering and acceleration / braking operations, achieving proactive and smooth control.

[0082] Scheduling parameters (Vehicle speed), matrix , and The parameters were obtained through offline linearization and data fitting methods at different vehicle speeds. Below, a sweep frequency signal is applied using high-precision vehicle dynamics simulation software (such as CarSim), and a system identification algorithm is used to obtain the most representative signal. and right Parameters of the linear model for dynamic response.

[0083] This invention employs Recursive Least Squares (RLS) as the core algorithm for system identification of the LPV model. Its advantages include: a "forgetting factor" to track time-varying parameters (adapting to vehicle speed). It features a variable LPV model; low computational cost (computation cost per iteration < 1ms), which can meet the real-time requirements of 10ms control cycle for embedded platforms; strong noise immunity, which can effectively handle measurement noise from IMU and wheel speed sensors (such as random noise of ±0.1° tilt angle).

[0084] Build a high-precision dynamic model in CarSim that matches the target vehicle model (inputting inherent parameters such as vehicle mass, wheelbase, tire lateral stiffness, and moment of inertia), and input the driver disturbance w (front wheel steering angle). Longitudinal acceleration A frequency sweep signal is applied as an excitation;

[0085] in, The sweep frequency range is [-10°, 10°], and the sweep frequency is 0.1~5Hz (covering daily steering to emergency lane change conditions). The sweep frequency range is [-0.6g, 0.6g], and the sweep frequency is 0.1~2Hz (covering emergency braking to full acceleration conditions); six typical vehicle speed points (10km / h, 30km / h, 50km / h, 80km / h, 120km / h, 150km / h) are selected, and the frequency is continuously swept for 300s at each speed, while the core status is collected simultaneously. The data is sampled at a frequency of 100Hz (matching the actual vehicle control cycle).

[0086] Data processing includes noise filtering: Kalman filtering is used to remove measurement noise in state x (retaining effective frequency components from 0.01 to 10 Hz and filtering out high-frequency vibration interference from the IMU); data normalization: input... With output Map to the [-1,1] interval to avoid identification errors caused by numerical overflow.

[0087] Parameter iterative estimation:

[0088] Discretize the LPV model into an identification standard form. ,in: Represents the output vector at time k (i.e. ); This represents the regression matrix at time k, which is composed of historical input and output data. The parameter vector to be identified; This represents the error term (modeling error and measurement noise).

[0089] A forgetting factor λ = 0.95 is introduced. λ is calibrated through 10 sets of identification experiments at different vehicle speeds to balance the weights of historical and new data and avoid parameter drift. The solution is obtained using the RLS iterative formula. :

[0090]

[0091]

[0092] When parameter iteration error When the parameters converge, save (A(ρ)), Bᵤ(ρ), and B at the current vehicle speed ρ. w (ρ) matrix.

[0093] Offline construction of a "vehicle speed-matrix parameter" lookup table (storing A and B values ​​for 6 typical vehicle speed points) u Bw (Parameters); During actual vehicle operation, the estimated vehicle speed is based on the current speed. (Wheel speed sensor fusion values), matrix parameters at any vehicle speed are calculated through linear interpolation (e.g., When the speed is 60km / h, interpolate the parameters for 50km / h and 80km / h to ensure that the LPV model adapts to changes in vehicle speed in real time.

[0094] MPC optimization problem construction and solution:

[0095] In each control cycle k, the MPC controller solves a finite-time domain problem. Internal optimization issues:

[0096] .

[0097] Constraints:

[0098] (System dynamics constraints)

[0099] (Actuator physical constraints)

[0100] (Actuator rate constraint)

[0101] Among them: U = [u(t), u(t+1), ..., u(t+ )] T It is the control sequence to be optimized; the output vector , For the output matrix, Reference trajectory , The control objective is clear and specific: to maintain the vehicle body in a level and stationary position.

[0102] and It is a positive definite weight matrix used to balance attitude stability accuracy and control energy consumption. These represent the roll angle weight and pitch angle weight, respectively. These represent the cost weights for the rear wheel steering angle, the front axle active stabilizer bar torque, and the rear axle active stabilizer bar torque, respectively.

[0103] P is the terminal weight matrix, which is obtained by solving the discrete-time algebraic Riccati equation to ensure closed-loop stability.

[0104] Actuator constraints:

[0105] Physical constraints (hard constraints): , ;

[0106] Rate constraints (soft constraints): [ (i.e., a maximum change of ±0.3° every 10ms period); .

[0107] Model update: based on the currently estimated vehicle speed The matrix of the prediction model is updated by lookup table interpolation. , , ;

[0108] Adaptive weighting: The weight matrices Q and R are dynamically interpolated and adjusted based on the confidence level of the identified working conditions;

[0109] Operating Condition A: High-speed cornering / emergency lane change (lateral stability is the primary factor)

[0110] Control objective: Prioritize suppressing roll angle;

[0111] MPC implementation: Increase Side tilt angle in the matrix weight ;

[0112] Typical weights: , ([50, 10, 10]).

[0113] Operating Condition B: Emergency Braking (Longitudinal Stability Dominant)

[0114] Control objective: Prioritize suppressing pitch angle ;

[0115] MPC implementation: Increase Pitch angle in the matrix weight ;

[0116] Typical weights: , ([20, 50, 50]).

[0117] Operating Condition C: Bumpy road surface (vertical comfort is the primary consideration)

[0118] Control objective: To limit actuator movement to improve comfort;

[0119] MPC implementation: Increase In the matrix The weights are used to limit its actions;

[0120] Typical weights: , ([10, 500, 500]).

[0121] Operating Condition D: Strong crosswind / single-sided adhesion (anti-interference stability)

[0122] Control objective: Quickly restore stable attitude;

[0123] MPC Implementation: Appropriately Increase Weighting reduces the cost of executor actions, allowing executors to respond quickly.

[0124] Typical weights: , ([30, 50, 50]).

[0125] The above weight values ​​are baseline settings under typical operating conditions. In actual online control, the cooperative controller can smoothly interpolate between the above baseline values ​​based on the continuous confidence level output by the operating condition identification module to achieve a smooth transition of the control strategy.

[0126] Optimization Solution: The embedded high-efficiency QP solver OSQP is invoked to solve the above optimization problem, and the first element of the optimal control sequence is taken as the current... Control commands are output to the actuator at any given time. In the next control cycle, the state estimate is updated based on the new measurements, and the optimization problem is solved again, achieving rolling optimization and feedback correction. The solution time is controlled within 3ms using a warm-start technique.

[0127] e. Command output:

[0128] The optimal control command, derived from the optimal control sequence, is executed by the active rear-wheel steering system and the active lateral stabilizer bar system to track the rear wheel steering angle and output the target anti-roll moment.

[0129] The following is a specific example of using the rear-wheel steering and active stabilizer bar coordinated control method of the present invention, including the selection and configuration of the coordinated controller hardware:

[0130] The core processor can be a high-performance automotive-grade microprocessor, such as the NXP S32G274A. This chip integrates multiple ARM Cortex-A53 / Cortex-M7 cores, supports ASIL-D functional safety level, and meets the real-time and reliability requirements of critical chassis systems.

[0131] The program is written in C / C++ and integrated into a software architecture that conforms to the AUTOSAR (AUTomotive Open System Architecture) standard. The cooperative control algorithm (MPC) runs as a standalone software component (SWC) and interacts with the underlying software modules and the vehicle bus via the RTE (Run-Time Environment).

[0132] The system runtime is set to 10ms, and the MPC prediction time domain is... (Corresponding to 100ms), control time domain It balances prediction accuracy with computational load.

[0133] Sensor accuracy and data fusion:

[0134] The accuracy requirements for key sensors such as IMU (Inertial Measurement Unit) are: yaw rate measurement accuracy better than 0.5° / s, and roll / pitch angle measurement accuracy better than 0.1°. The accuracy requirement for wheel speed sensors is: speed measurement error less than 0.1 km / h.

[0135] Actuator Interface and Constraints:

[0136] The maximum rotation angle range of the ARS actuator is set to ±5°, and the maximum actuation speed is set to 30° / s;

[0137] The maximum output torque range of the AAS actuator is set to ±1500 Nm;

[0138] The collaborative controller sends control commands to the ECUs of ARS and AAS via the CAN FD bus.

[0139] Several control examples of the coordinated control method implemented based on the rear-wheel steering and active stabilizer bar coordinated control system of this embodiment are as follows:

[0140] Example 1: Emergency double lane change at 100km / h (lateral stability)

[0141] Test environment: road surface adhesion coefficient 0.8 (dry asphalt road), vehicle fully loaded, initial speed 100km / h.

[0142] Driver's operation: The driver quickly turns the steering wheel to make an emergency lane change (steering wheel angular velocity 180deg / s, lateral acceleration 0.5g).

[0143] System response:

[0144] 1) Operating condition identification: The system identifies the operating condition as "emergency lane change".

[0145] 2) MPC decision: Call the emergency lane change weight and solve for the optimal control quantity: the rear wheels rotate 0.8° in the same direction and the front axle active stabilizer bar outputs 800 Nm anti-roll torque.

[0146] 3) Synergistic effect: The angle of the ARS reduces the roll moment at the source; the AAS accurately compensates for the remaining roll.

[0147] 4) Effect: The body roll angle is controlled within 2° (traditional solutions >5°), and the lateral displacement deviation of the vehicle is significantly reduced.

[0148] Example 2: Emergency braking at 100 km / h (longitudinal stability)

[0149] Test environment: initial braking speed 100km / h, road surface μ=0.8, brake master cylinder pressure 100bar.

[0150] Driver's operation: The driver slams on the brake pedal (opening >90%).

[0151] System response:

[0152] 1) Operating condition identification: The system identifies the operating condition as "emergency braking";

[0153] 2) MPC decision: Increase the pitch angle weight and solve for the control quantity: rear wheel reverse rotation -0.3°, front axle AAS output -500Nm, rear axle AAS output 400Nm;

[0154] 3) Anti-pitch mechanism: The reverse steering angle of the rear wheels changes the distribution of the lateral force of the rear wheels, generating a pitching moment that lifts the front of the car, which works in conjunction with the force couple of the AAS to counteract the braking pitching moment.

[0155] 4) Effect: The vehicle pitch angle was reduced from 4.2° in the traditional solution to 1.2° (a reduction of 71%).

[0156] Example 3: High-speed driving over a road surface with continuous speed bumps (vertical comfort)

[0157] Test environment: speed bump height 5cm, spacing 1.5m, vehicle speed 80km / h.

[0158] System response:

[0159] 1) Working condition identification: The system identifies the "bumpy road surface" working condition by high-frequency fluctuations in wheel speed;

[0160] 2) MPC decision: Increase the torque weight of AAS to limit its action; ARS performs ±0.2° high-frequency micro-motion to assist in vibration absorption;

[0161] 3) Effects: AAS equivalent stiffness is reduced, wheel ground contact is improved by 30%, and vertical impact acceleration inside the vehicle is reduced by 30%.

[0162] Simulations were performed using both traditional independent control and the method of this invention under the same emergency lane change conditions. The results are as follows: Figure 3 As shown, the beneficial effects and underlying mechanisms of the present invention are quantitatively demonstrated, specifically as follows:

[0163] 1) Comparison of overall performance indicators (corresponding) Figure 3 (d: vehicle roll angle response)

[0164] Traditional control strategy: The body roll angle response is severe, with the maximum peak exceeding ±10°, indicating serious instability of the body posture and low handling limits.

[0165] The collaborative control strategy of this invention significantly suppresses the vehicle body roll angle, with the maximum peak value controlled within ±2°.

[0166] Quantitative conclusion: Under the test conditions, the present invention reduced the peak body roll angle by more than 80%, which greatly improved the lateral stability and safety of the vehicle.

[0167] 2) Actuator coordination mechanism analysis (corresponding to) Figure 3 (a, b, c)

[0168] The performance improvement of this invention stems from the model-predicted optimized torque distribution between the Active Rear Steering (ARS) and Active Stabilizer Bar (AAS) systems:

[0169] ARS role analysis ( Figure 3 (a)

[0170] In traditional control, the rear wheel steering angle movement of ARS is small and does not have an effective effect on vehicle body roll.

[0171] In this invention, the cooperative controller calculates the optimal rear wheel steering angle command, causing it to generate a yaw moment and lateral force opposite to the roll direction, thus actively counteracting part of the roll moment from its dynamic origin. This demonstrates that the ARS plays a feedforward dominant control role in the cooperative control.

[0172] AAS effect analysis ( Figure 3 (b, c):

[0173] In traditional control, the AAS system needs to independently provide all anti-roll moments, resulting in front and rear axle moments ( The output has a high peak value and a large gradient.

[0174] In this invention, since the ARS system already undertakes the basic anti-roll task, the peak torque output required by the AAS system is significantly reduced, and the dynamic changes are smoother. This demonstrates that the AAS primarily plays a role in feedback fine compensation in cooperative control.

[0175] 3) Summary of the core beneficial effects of collaborative control

[0176] Global performance optimization: Through integrated MPC solution, the optimal allocation of ARS and AAS control variables is achieved, solving the control island problem.

[0177] Energy efficiency optimization: While achieving better attitude stability, the torque requirements of the AAS system are reduced. (Reducing) helps to extend actuator life and reduce energy consumption.

[0178] Balancing comfort and handling: The active intervention of ARS reduces reliance on AAS, avoiding the comfort degradation caused by excessive "stiffening" of the suspension in pursuit of stability in traditional solutions. This invention achieves a balance between high stability and high comfort.

Claims

1. A method for coordinated control of rear-wheel steering and active stabilizer bar, characterized in that, include: Collect vehicle status signals and driver commands; Extended Kalman filtering is used to estimate the current vehicle state vector, which includes roll angle, roll rate, pitch angle, and pitch rate, based on the vehicle state signal. The current operating condition of the vehicle is determined by multiple conditions based on vehicle status signals, driver commands, and a pre-set rule base. Model predictive control is performed using a linear parametric time-varying model as the predictive model for vehicle state: Based on the current vehicle state vector, the objective function is solved under the system dynamics constraints, actuator physical constraints, and rate constraints of the linear parametric time-varying model to obtain the optimal control sequence for the rear wheel steering angle, front axle active stabilizer bar moment, and rear axle active stabilizer bar moment. The objective function is to minimize the sum of attitude deviation cost, actuator action cost, and terminal state deviation cost within a finite time domain. The matrix parameters of the linear parametric time-varying model are adjusted based on vehicle speed, and the attitude deviation cost and actuator action cost are adjusted based on the current operating condition of the vehicle. The linear parametric time-varying model is expressed as follows: ， in, This represents the vehicle's state vector. For control vectors, For driver instructions, , and For matrix parameters, Indicates time; The optimal control command obtained based on the optimal control sequence is executed by the active rear wheel steering system and the active lateral stabilizer bar system to perform rear wheel steering angle tracking and output the target anti-roll moment.

2. The rear-wheel steering and active stabilizer bar coordinated control method according to claim 1, characterized in that, The vehicle's operating modes include high-speed cornering / emergency lane change, emergency braking, bumpy road surface, and strong crosswind / single-sided adhesion. The attitude deviation cost and actuator action cost are adjusted based on the vehicle's current operating conditions by setting calculation weight benchmark values ​​for attitude deviation cost and actuator action cost for each operating condition. Then, based on the continuous confidence of the operating mode to which the vehicle's current operating condition belongs, output by the fuzzy logic algorithm, the calculation weight benchmark values ​​for attitude deviation cost and actuator action cost are smoothly interpolated to obtain the calculation weights for attitude deviation cost and actuator action cost.

3. The rear-wheel steering and active stabilizer bar coordinated control method according to claim 1, characterized in that, The objective function is expressed as follows: ， Where U is the control sequence to be optimized. Representing a finite time domain, , The vehicle's roll angle. The vehicle's pitch angle. , The weights for calculating the attitude deviation cost are... For control vectors, The weights for calculating the cost of actuator actions. Let be the vehicle's state vector. , The weights for calculating the terminal state deviation cost.

4. The rear-wheel steering and active stabilizer bar coordinated control method according to claim 2, characterized in that, The weighting benchmark for calculating the attitude deviation cost under high-speed cornering / emergency lane change conditions is... The baseline value for calculating the actuator action cost is... ([50, 10, 10]); The baseline weight for calculating the attitude deviation cost under emergency braking conditions is [50, 10, 10]. The baseline value for calculating the actuator action cost is... ([20, 50, 50]); The benchmark weight for calculating the attitude deviation cost under bumpy road conditions is (20, 50, 50). The baseline value for calculating the actuator action cost is... ([10, 500, 500]); The baseline weight for calculating the attitude deviation cost under strong crosswind / single-sided adhesion conditions is [value missing]. The baseline value for calculating the actuator action cost is... ([30,50, 50]).

5. The rear-wheel steering and active stabilizer bar coordinated control method according to claim 1, characterized in that, The matrix parameters of the linear parameter time-varying model are adjusted based on vehicle speed by constructing a vehicle speed-matrix parameter lookup table offline, and then calculating the matrix parameters at the current vehicle speed through linear interpolation based on the matrix parameter values ​​corresponding to a specific vehicle speed in the lookup table.

6. The rear-wheel steering and active stabilizer bar coordinated control method according to claim 5, characterized in that, The offline construction of the vehicle speed-matrix parameter lookup table includes the following steps: A linear parameter time-varying model matching the target vehicle model is built in the simulation software, and a sweep frequency signal is applied as an excitation to the driver's command; Noise filtering is used to remove measurement noise of the vehicle state obtained from a more dynamic model, and the vehicle state and driver commands are normalized. The linear parameter time-varying model is discretized into an identification standard form, and then the matrix parameters of the corresponding linear parameter time-varying model at the current vehicle speed are obtained by using the recursive least squares method until the parameters converge.

7. A rear-wheel steering and active stabilizer bar coordinated control system, characterized in that, include: The data acquisition module is used to collect vehicle status signals and driver commands; The state estimation module is used to estimate the current vehicle state vector, including roll angle, roll rate, pitch angle, and pitch rate, based on the vehicle state signal using extended Kalman filtering. The operating condition estimation module is used to determine the current operating condition of the vehicle based on vehicle status signals, driver commands, and a pre-set rule base through multiple conditions. The predictive control module is used to perform model predictive control based on a linear parametric time-varying model as the predictive model for vehicle state. Based on the current vehicle state vector, under the system dynamics constraints, actuator physical constraints, and rate constraints of the linear parametric time-varying model, it solves the objective function to obtain the optimal control sequence for the rear wheel steering angle, front axle active stabilizer bar moment, and rear axle active stabilizer bar moment. The objective function minimizes the sum of attitude deviation cost, actuator action cost, and terminal state deviation cost within a finite time domain. The matrix parameters of the linear parametric time-varying model are adjusted based on vehicle speed, and the attitude deviation cost and actuator action cost are adjusted based on the vehicle's current operating conditions. The linear parametric time-varying model is expressed as follows: ， in, This represents the vehicle's state vector. For control vectors, For driver instructions, , and For matrix parameters, Indicates time; as well as, The execution module includes an active rear-wheel steering system and an active lateral stabilizer bar system, which are used to execute the optimal control command obtained based on the optimal control sequence, perform rear wheel steering angle tracking, and output the target anti-roll moment.

Images