Multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope

Abstract

The application discloses a multi-actuator collaborative chassis stability control method based on a four-dimensional dynamic safety envelope and belongs to the technical field of vehicle dynamics control. The application solves the technical problem that the existing vehicle stability control method is easy to break through the vehicle stability domain and significantly increases the instability risk under extreme working conditions. Vehicle state estimation is performed by collecting vehicle running state information, actuator feedback information and road environment information, and a four-dimensional dynamic safety envelope is constructed based on the estimation result. The safety margin corresponding to each envelope is calculated, normalized and fused into a unified safety margin, the vehicle state is regionally divided, and the region where the vehicle is located and the region membership degree are obtained. The control weight of the control target is continuously adjusted according to the region where the vehicle is located and the region membership degree. Based on the control weight, a hierarchical optimization method is used to solve the multi-actuator collaborative control quantity, and the multi-actuator collaborative control quantity is output to the corresponding actuator of the vehicle. The application is used for realizing the multi-actuator collaborative chassis stability control under extreme working conditions.

Description

Technical Field

[0001] This invention relates to the field of vehicle dynamics control technology, specifically to a multi-actuator collaborative chassis stability control method based on a four-dimensional dynamic safety envelope. Background Technology

[0002] With the development of electrification, intelligence and drive-by-wire technologies in new energy vehicles, vehicle chassis control systems are gradually evolving from traditional single-actuator control to multi-actuator collaborative control.

[0003] Traditional methods typically prioritize minimizing trajectory tracking error in vehicle control. However, under extreme conditions such as high-speed steering, emergency obstacle avoidance, driving on low-traction surfaces, or forced dynamic steering coupling, significant coupling relationships exist between the vehicle's longitudinal, lateral, yaw, and vertical dynamics. Vehicle yaw rate, sideslip angle, tire longitudinal force, tire lateral force, tire normal load, and actuator output capability interact with each other. When the vehicle approaches its traction limit, if the control system continues to prioritize minimizing trajectory tracking error, it may lead to a rapid increase in tire utilization, causing yaw rate and sideslip angle to continuously deviate from safe ranges, resulting in instability phenomena such as sideslip, fishtailing, understeer, or oversteer. Existing technologies still have significant limitations in handling the dynamic coupling relationships in various directions, making it easy to exceed the vehicle's stability domain and significantly increasing the risk of instability.

[0004] In summary, under extreme operating conditions, existing vehicle stability control methods suffer from the technical problem that the vehicle stability domain is easily breached, significantly increasing the risk of instability. Summary of the Invention

[0005] This invention solves the technical problem that existing vehicle stability control methods are prone to exceeding the vehicle stability domain and significantly increasing the risk of instability under extreme operating conditions.

[0006] The multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope described in this invention includes the following steps: Step 1: Collect vehicle operating status information, actuator feedback information, and road environment information; Step 2: Estimate the vehicle state based on vehicle operating status information, actuator feedback information and road environment information, and construct a four-dimensional dynamic safety envelope based on the estimation results; Step 3: Calculate the safety margin corresponding to each envelope in the four-dimensional dynamic safety envelope, and normalize and merge the safety margins corresponding to each envelope into a unified safety margin. Step 4: Based on a unified safety margin, divide the vehicle status into regions, and obtain the region where the vehicle is located and the region membership degree based on the division results. Step 5: Continuously adjust the control weights of the control target based on the vehicle's location and the region's membership degree; Step 6: Based on the control weights, use a hierarchical optimization method to solve the multi-actuator collaborative control quantity, and output the multi-actuator collaborative control quantity to the corresponding actuator of the vehicle to complete the chassis stability control of the current control cycle.

[0007] Furthermore, in one embodiment of the present invention, the vehicle state estimation in step 2 includes the vehicle's longitudinal speed, lateral speed, yaw rate, center of gravity sideslip angle, road surface adhesion coefficient, tire utilization rate, normal load of each wheel, and remaining margin of steering, driving, braking and active suspension actuators.

[0008] Furthermore, in one embodiment of the present invention, the four-dimensional dynamic safety envelope in step 2 specifically refers to: ; in, For a stable lateral envelope, For lateral stability envelope, For tire utilization envelope, For the normal load transfer envelope, This is the actuator margin envelope.

[0009] Furthermore, in one embodiment of the present invention, the yaw stabilizing envelope specifically comprises: ; in, The initial upper boundary of the yaw rate. This is the vehicle state vector. The vehicle's yaw rate; The lateral stabilizing envelope is specifically: ; in, The sideslip angle is the angle at the vehicle's center of gravity. The upper boundary of the centroid sideslip angle; The tire utilization rate envelope is specifically as follows: ; in, For the first The overall utilization rate of each wheel's tires This represents the upper boundary of tire utilization. The normal load transfer envelope includes normal load constraints for each round, lateral load transfer rate constraints, and longitudinal load transfer rate constraints. The actuator margin envelope includes steering system constraints, drive system constraints, braking system constraints, and active suspension system constraints.

[0010] Furthermore, in one embodiment of the present invention, the four-dimensional dynamic safety envelope is corrected by using a control invariant set, and the corrected four-dimensional dynamic safety envelope is scaled to obtain the final four-dimensional dynamic safety envelope.

[0011] Furthermore, in one embodiment of the present invention, step 4, based on a unified safety margin, involves dividing the vehicle status into regions, specifically as follows: Set stability threshold and critical threshold ,when At that time, the vehicle was in a stable zone, when At that time, the vehicle was in the critical zone, when At that time, the vehicle was in an unstable zone, among which, To standardize safety margins.

[0012] Furthermore, in one embodiment of the present invention, the region division is determined using hysteresis logic or a fuzzy state machine.

[0013] Furthermore, in one embodiment of the present invention, the step 5 of continuously adjusting the control weight of the control target based on the region where the vehicle is located and the region membership degree specifically includes: When the membership degree of the stable region increases, the weights of the trajectory tracking target and the actuator smoothness target are increased; When the membership degree of the critical region increases, the weights of the yaw stability target, the side slip stability target, the load transfer target, and the tire utilization control target are increased. When the membership degree of the unstable region increases, the weight of the safe recovery target is increased, and the weight of the trajectory tracking target is decreased.

[0014] Furthermore, in one embodiment of the present invention, the step 6 of solving the multi-actuator cooperative control quantity using a hierarchical optimization method specifically involves: The upper layer is based on the vehicle dynamics model and the four-dimensional dynamic safety envelope, with the control weight as the weighting coefficient of the objective function. The nonlinear model predicts and controls the front wheel steering angle, the desired longitudinal resultant force, the desired additional yaw moment, and the desired roll control moment. The lower layer distributes the desired longitudinal resultant force and the desired additional yaw moment into torque commands for each wheel drive motor and braking torque commands for each wheel, and distributes the desired roll control moment into power commands for each wheel active suspension.

[0015] Furthermore, in one embodiment of the present invention, the method further includes: Step 7: Obtain actuator feedback and vehicle response, update vehicle state estimation, four-dimensional dynamic safety envelope and control target based on actuator feedback and vehicle response, and enter the next control cycle.

[0016] This invention solves the technical problem that existing vehicle stability control methods, under extreme operating conditions, are prone to exceeding the vehicle's stability domain, significantly increasing the risk of instability. Specific beneficial effects of this invention include: 1. This invention proposes a multi-actuator cooperative chassis stability control method based on a four-dimensional dynamic safety envelope. By constructing a four-dimensional dynamic safety envelope, attitude stability, tire adhesion utilization, load transfer, and actuator margin are uniformly incorporated into the safety judgment system, enabling a more comprehensive identification of vehicle instability risks under extreme conditions. A hierarchical optimization allocation method is adopted to integrate the front-wheel steering system, four-wheel drive system, four-wheel braking system, and active suspension system into a unified coordination framework. This fully leverages the advantages of multi-actuator cooperative control, improving vehicle stability, trajectory tracking performance, and control robustness under complex conditions such as low adhesion, high-speed obstacle avoidance, and extreme steering. This solves the technical problem of easily breaching the vehicle's stability domain and significantly increasing instability risk. 2. This invention proposes a multi-actuator cooperative chassis stability control method based on a four-dimensional dynamic safety envelope. The method uses a mechanism model to generate the initial envelope, corrects the control invariant set, and scales the data online to construct the dynamic safety envelope. This allows the safety boundary to change in real time according to the road adhesion coefficient, vehicle speed, driving intention, and actuator working state, thereby improving the adaptability and robustness of the safety boundary. 3. This invention proposes a multi-actuator collaborative chassis stability control method based on a four-dimensional dynamic safety envelope. It achieves the division of stable, critical, and unstable regions by using a unified safety margin index or fuzzy state machine, and suppresses region switching jitter by using hysteresis band or small hysteresis logic, thereby improving the smoothness of control mode switching. 4. This invention proposes a multi-actuator cooperative chassis stability control method based on a four-dimensional dynamic safety envelope. The method adaptively generates control target priorities according to the region where the vehicle is located, so that the vehicle prioritizes trajectory tracking and ride comfort in the stable region, prioritizes yaw stability and lateral stability in the critical region, and prioritizes safety recovery and return to the dynamic safety envelope in the unstable region. This invention is particularly applicable to distributed drive electric vehicles equipped with front-wheel steering, four-wheel independent drive, four-wheel independent braking, and active suspension. It can also be applied to four-wheel independent drive vehicles, drive-by-wire chassis vehicles, autonomous vehicles, and intelligent electric vehicles with multi-actuator cooperative control capabilities. This method can be used for chassis cooperative stability control under complex conditions such as low-traction road surface steering, high-speed lane changing, emergency obstacle avoidance, forced dynamic steering coupling, extreme cornering, and vehicles approaching tire adhesion limits. Attached Figure Description

[0017] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1This is a flowchart of the multi-actuator cooperative chassis stability control method described in Implementation Method 1; Figure 2 This is a structural diagram of the vehicle multi-actuator cooperative chassis control system described in Implementation Method 1; Figure 3 This is a flowchart of the four-dimensional dynamic security envelope construction described in Implementation Method 3; Figure 4 This is a schematic diagram illustrating the state division of the stable region, critical region, and unstable region as described in Implementation Method 4; Figure 5 This is the flowchart of the fuzzy state machine risk determination described in Implementation Method 4; Figure 6 This is the region-driven hierarchical optimization allocation structure diagram described in Implementation Method Six; Figure 7 It is a vehicle configuration diagram with steering, driving, braking and active suspension as described in Embodiment 7. Detailed Implementation

[0018] Various embodiments of the present invention will now be clearly and completely described with reference to the accompanying drawings. The embodiments described with reference to the drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0019] Implementation Method 1: Existing technologies struggle to achieve dynamic perception and accurate assessment of a vehicle's multi-dimensional safety status under extreme operating conditions. Furthermore, the control strategy lacks adaptive adjustment capabilities based on risk status, and there is insufficient coordination among actuators, leading to the vehicle's stability domain being easily breached and a significant increase in the risk of instability.

[0020] To address the aforementioned technical problems, this embodiment proposes a multi-actuator cooperative chassis stability control method based on a four-dimensional dynamic safety envelope, such as... Figure 1 As shown, the specific steps include: Step 1: Collect vehicle operating status information, actuator feedback information, and road environment information; Step 1 is used to obtain the basic data required for vehicle chassis control. For example... Figure 2 As shown, the information is collected by sensor group 2, which includes inertial measurement unit 21, wheel speed sensor 22, steering angle sensor 23, suspension displacement sensor 24, brake pressure sensor 25, drive torque feedback unit 26 and environmental perception unit 27.

[0021] Step 11: Collect vehicle inertial state information The inertial measurement unit 21 is positioned near the vehicle's center of gravity, or equivalently at the vehicle's center of gravity after coordinate transformation. The inertial measurement unit 21 is used to collect the vehicle's longitudinal acceleration. lateral acceleration yaw rate yaw rate And the vehicle's attitude angle.

[0022] The acquired inertial signal is input to parameter estimation module 3 after undergoing zero-bias compensation, gravity component compensation, and low-pass filtering. Preferably, the sampling frequency of the inertial measurement unit 21 is 100Hz to 200Hz.

[0023] Step 12, collect wheel motion status information Wheel speed sensors 22 are respectively arranged at the four wheels to collect the wheel speed of the left front wheel. Right front wheel speed Left rear wheel speed and the speed of the right rear wheel . No. The linear velocity of the wheel flange Represented as: ; in, For the first The linear velocity of the wheel flange. The effective rolling radius of the tire. For the first The wheel speed of each wheel.

[0024] Step 13, Collect steering status information Steering angle sensor 23 is used to collect the front wheel steering angle. and the rate of change of front wheel steering angle When the front wheel angle sensor 23 and the steering wheel angle sensor are both present, the steering wheel angle can be converted into the front wheel angle according to the steering transmission ratio, and the accuracy of the front wheel angle estimation can be improved through sensor fusion.

[0025] Step 14, Collect suspension status information Suspension displacement sensor 24 is used to collect the first... The relative displacement of the suspension corresponding to each wheel Its definition is: ; in, For the first Each wheel corresponds to a vertical lateral displacement of the vehicle body. For the first Each wheel corresponds to the vertical displacement of the unsprung mass.

[0026] Step 15: Collect actuator feedback information Brake pressure sensor 25 is used to collect the first... Braking torque feedback value of each wheel Or braking pressure feedback value. The drive torque feedback unit 26 is used to collect the first... Torque of the drive motor for each wheel Motor speed and motor temperature. Active suspension system 94 feedback... An active suspension system as the power source Suspension actuator displacement and suspension actuator speed.

[0027] Step 16: Collect road environment information and driving intention information. The environmental perception unit 27 includes at least one of a camera, millimeter-wave radar, lidar, high-precision map, and road surface recognition unit, used to acquire road curvature. The target trajectory, distance to obstacles, road slope, prior information on road surface adhesion, and driving intention information.

[0028] driving intent intensity It can be determined comprehensively based on the reference road curvature, front wheel steering angle change rate, reference lateral acceleration, and obstacle risk: ; in, This is the normalized amplitude limiting function; , , , These are the weighting coefficients for road curvature, front wheel steering angle change rate, reference lateral acceleration, and obstacle risk, respectively. For the first Reference lateral acceleration at each control moment; For the first Obstacle risk indicators at each control moment.

[0029] Step 17, Signal Synchronization and Preprocessing The signals from all sensors are synchronized according to a unified timestamp, and the sampling period is controlled to be [missing information]. Preferably, The timeout can be set from 0.005s to 0.02s. For signals with high noise levels, preprocessing is performed using low-pass filtering, Kalman filtering, or moving average filtering. For sensor signals that are lost for a short time, compensation is performed using the estimated value or state prediction value from the previous moment.

[0030] Step 2: Estimate the vehicle state based on vehicle operating status information, actuator feedback information and road environment information, and construct a four-dimensional dynamic safety envelope based on the estimation results; Step 3: Calculate the safety margin corresponding to each envelope in the four-dimensional dynamic safety envelope, and normalize and merge the safety margins corresponding to each envelope into a unified safety margin. Step 4: Based on a unified safety margin, divide the vehicle status into regions, and obtain the region where the vehicle is located and the region membership degree based on the division results. Step 5: Continuously adjust the control weights of the control target based on the vehicle's location and the region's membership degree; Step 6: Based on the control weights, use a hierarchical optimization method to solve the multi-actuator collaborative control quantity, and output the multi-actuator collaborative control quantity to the corresponding actuator of the vehicle to complete the chassis stability control of the current control cycle.

[0031] This implementation obtains a unified safety margin by normalizing the safety margins of each dimension and combines it with the dynamic determination of the region where the vehicle is located; it continuously adjusts the control weights by region membership; and it solves the multi-actuator collaborative control quantity from the upper layer by using a hierarchical optimization method, and then distributes it to the corresponding actuators of the vehicle, thereby completing the multi-actuator collaborative chassis stability control. This solves the problems of insufficient safety boundary dimensions, weak dynamic adaptability, easy jitter in region determination, insufficient multi-actuator collaboration, and fixed control target priority in existing vehicle stability control methods, and reduces the risk of the vehicle stability domain being easily breached and unstable.

[0032] Implementation Method 2: The difference between this implementation method and Implementation Method 1 is that the vehicle state estimation in step 2 includes the vehicle's longitudinal speed, lateral speed, yaw rate, center of gravity sideslip angle, road surface adhesion coefficient, tire utilization rate, normal load of each wheel, and remaining margin of steering, driving, braking, and active suspension actuators.

[0033] This implementation method estimates vehicle state parameters online, specifically including: Step 21, estimate the vehicle's longitudinal velocity Under low slip conditions, the vehicle's longitudinal speed It can be obtained by weighted fusion of the linear velocities of the four wheel flanges. The linear velocity of the wheel flange Represented as: ; in, For the first The first control moment The linear velocity of the wheel flange; The effective rolling radius of the tire. For the first The first control moment The wheel speed of each wheel; ,in, The left front wheel, The right front wheel, It is the left rear wheel. It is the right rear wheel.

[0034] Under low slip conditions, the initial estimate of the vehicle's longitudinal velocity can be expressed as: ; in, For the first The initial estimate of the vehicle's longitudinal velocity at each control moment; For the first The initial wheel speeds of each wheel are fused with weights, and the following conditions are met: ; In one implementation, if each wheel is in a low-slip state, the following can be taken: ; In another implementation, the wheel slip state, wheel speed signal reliability, or drive / brake state can also be considered. Configure adaptive settings.

[0035] When a vehicle experiences driving slip or braking slip, a longitudinal slip ratio correction mechanism is introduced to reduce the impact of slip wheel speed on the estimation of the vehicle's longitudinal speed. During the nth iteration, the 1st The estimated longitudinal slip ratio of each wheel is defined as: ; in, For the first The first control moment The result of the iteration is the first Estimated longitudinal slip ratio of each wheel; For the k-th control time, the first... The estimated longitudinal velocity of the vehicle obtained in the next iteration; To prevent positive numbers with a denominator of zero; This represents the number of iterations.

[0036] To reduce the impact of wheels with large slippage on the estimation of vehicle longitudinal velocity, the wheel speed fusion weights are updated based on the longitudinal slip ratio. During the nth iteration, the 1st The fusion weight of each wheel can be expressed as: ; in, For the first During the nth iteration The wheel speed fusion weight of each wheel; To prevent the weight calculation from being a positive number with a denominator of zero; The wheel index is used. As shown in the formula above, the greater the longitudinal slip ratio of the wheel, the smaller its corresponding wheel speed fusion weight.

[0037] Based on the updated fusion weights and longitudinal slip ratio, the vehicle's longitudinal speed is iteratively corrected: ; in, For the first The first control moment The estimated longitudinal velocity of the vehicle obtained in the next iteration.

[0038] To ensure the real-time performance and stability of the longitudinal velocity estimation, iterative convergence conditions and a maximum number of iterations are set. When the conditions are met... or satisfy When the time limit is reached, the iteration is stopped, and the current iteration result is used as the estimated longitudinal velocity of the vehicle at the k-th control time. ; in, The convergence threshold for estimating the longitudinal velocity of the vehicle; This represents the maximum number of iterations. For the first The estimated longitudinal speed of the vehicle at each control moment.

[0039] The above iterative estimation method can quickly obtain the vehicle's longitudinal velocity estimate under low slip conditions; under driving slip or braking slip conditions, it can reduce the impact of abnormal wheel speed of slipping wheels on the vehicle's longitudinal velocity estimate, and improve the accuracy of subsequent centroid sideslip angle estimation, road surface adhesion coefficient estimation and four-dimensional dynamic safety envelope construction.

[0040] Step 22, estimate the lateral velocity and the sideslip angle of the center of mass. The vehicle's planar lateral dynamic equations are: ; Based on the above equation, the discretized transverse velocity can be predicted as follows: ; in, For the first The estimated lateral velocity at each control moment. For the first The first control time estimate Lateral tire force of each wheel.

[0041] The estimated centroid sideslip angle is: ; Step 23, estimate the road surface adhesion coefficient This embodiment preferably uses an unscented Kalman filter to estimate the road surface adhesion coefficient. The estimator state vector is defined as: ; in, For the longitudinal speed of the vehicle, For lateral velocity, The sideslip angle is the angle of the center of mass. The road surface adhesion coefficient, For the normal load of each wheel, This refers to the overall utilization rate of each tire.

[0042] Define the input vector of the estimator as: ; The state equation of the estimator is: ; in, This is a nonlinear state transition function constructed based on vehicle dynamics and a tire model. This is process noise.

[0043] Define the measurement vector as follows: ; The measurement equation is: ; in, For measurement functions, For measuring noise.

[0044] The unscented Kalman filter obtains the adhesion coefficient estimate by generating Sigma points, propagating Sigma points, calculating the predicted mean and predicted covariance, constructing measurement predictions, and correcting them based on actual measurements. To ensure that the estimation results meet physical requirements, constraints are imposed on the estimated adhesion coefficient: ; in, This is the lower limit of the road surface adhesion coefficient. This represents the upper limit of the road surface adhesion coefficient.

[0045] Step 24, estimate the normal load on each wheel. No. The normal load on each wheel is composed of static normal load, longitudinal load transfer, lateral load transfer, and load changes caused by the operation of the active suspension. ; in, The change in normal load caused by longitudinal acceleration. This refers to the change in normal load caused by lateral acceleration and vehicle roll. This refers to the change in normal load caused by the active suspension actuation.

[0046] Taking a four-wheeled vehicle as an example, the estimated normal load on each wheel can be expressed as: ; ; ; ; The load change caused by the active suspension actuation is: ; in, For the first The equivalent stiffness of the suspension. For the first The equivalent damping of the suspension. For the first The relative speed of the suspension.

[0047] Step 25, estimate the overall tire utilization rate No. The overall tire utilization rate of each wheel is defined as: ; in, For the first The first wheel in the Tire utilization rate at each control moment.

[0048] The longitudinal force of the tire can be estimated from the driving and braking torques: ; Tire lateral force can be estimated from the tire slip angle and the normalized tire force function: ; in, To normalize the lateral tire force function, This refers to the tire slip angle.

[0049] Implementation Method 3: The difference between this implementation method and Implementation Method 1 is that in this implementation method, the four-dimensional dynamic safety envelope consists of an attitude stability dimension, a tire adhesion utilization dimension, a load transfer dimension, and an actuator margin dimension. The attitude stability dimension includes yaw stability envelope and side slip stability envelope; the tire adhesion utilization dimension includes the tire utilization rate envelope for each wheel; the load transfer dimension includes the upper and lower limits of the normal load for each wheel and the longitudinal / lateral load transfer rate envelope; and the actuator margin dimension includes the amplitude margin and rate of change margin of the steering, drive, braking, and active suspension actuators. The construction process of the four-dimensional dynamic safety envelope is as follows: Figure 3 As shown.

[0050] Existing vehicle stability control methods typically use yaw rate and sideslip angle as the main evaluation indicators, and determine whether the vehicle is unstable by constructing a two-dimensional stability domain consisting of yaw rate boundary, sideslip angle boundary, or yaw rate-slip angle. While these methods can reflect the basic stability of the vehicle's planar motion, they are insufficient to fully describe the vehicle's true safety margin under extreme conditions.

[0051] Specifically, when a vehicle is in a state of coupled driving, braking, and steering, the longitudinal and lateral forces of the tires jointly occupy the adhesion circle, and the overall tire utilization rate may reach a dangerous state before the yaw rate or the sideslip angle. If the control system only judges stability based on the yaw rate and sideslip angle, it may miss the best time to intervene.

[0052] Meanwhile, vehicles experience significant lateral load transfer during high-speed cornering or emergency obstacle avoidance, and significant longitudinal load transfer during forced movement or rapid acceleration. Changes in normal load directly affect tire grip, and active suspension systems can actively alter the distribution of normal load on the tires. Therefore, if existing safety boundaries do not incorporate normal load transfer into their judgment system, it will be difficult to fully leverage the role of active suspension in chassis stability control.

[0053] Furthermore, steering, drive, braking, and suspension actuators all have amplitude and rate-of-change limitations. When the actuators approach saturation, even if the theoretical control input can bring the vehicle back to a stable state, the actual actuators may fail to achieve the expected control effect due to insufficient remaining margin. Therefore, actuator margin should also be considered an important component of the vehicle's safety boundary.

[0054] To address the aforementioned technical problems, this embodiment proposes a dynamic safety envelope construction method that simultaneously considers yaw stability, side slip stability, tire utilization rate, normal load transfer, and actuator margin. Specifically: ; in, For a stable lateral envelope, For lateral stability envelope, For tire utilization envelope, For the normal load transfer envelope, This is the actuator margin envelope.

[0055] This means that a vehicle is considered to be within the four-dimensional dynamic safety envelope only when it simultaneously satisfies the constraints of attitude stability, tire adhesion utilization, load transfer, and actuator margin. Among them, the yaw stability envelope and the sideslip stability envelope together constitute the attitude stability dimension, the tire utilization envelope constitutes the tire adhesion utilization dimension, the normal load transfer envelope constitutes the load transfer dimension, and the actuator margin envelope constitutes the actuator margin dimension.

[0056] The yaw stabilizing envelope is specifically as follows: Yaw stabilization envelope is used to constrain the vehicle's yaw rate. The initial boundary for the yaw rate is: ; in, The initial upper boundary of the yaw rate. For the lateral safety factor, and .

[0057] Combining the reference road curvature and front wheel steering angle, the yaw limit is further expressed as: ; in, The yaw boundary is obtained based on the mechanism model. This is the curvature correction factor. This is the corner correction factor. This is the boundary compensation value for low-speed yaw.

[0058] The yaw stability envelope is: ; in, The initial upper boundary of the yaw rate. This is the vehicle state vector. This represents the vehicle's yaw rate; the formula means that the current yaw rate cannot exceed the dynamic upper limit.

[0059] The lateral stabilizing envelope is specifically: Lateral stability envelope is used to constrain the vehicle's center of gravity sideslip angle. The initial boundary of the sideslip angle is: ; in, The initial upper boundary of the centroid sideslip angle. For the lateral deviation safety factor, This is the equivalent lateral stiffness of the entire vehicle.

[0060] Considering the influence of vehicle speed and driving intention, the sideslip angle boundary is: ; in, ; ; in, This is the speed scaling factor. This is the reference value for the lateral boundary velocity. and These are the lower and upper limits of the speed scaling factor, respectively. The scaling factor for driving intention. The contraction coefficient of the driving intention with respect to the lateral deviation boundary.

[0061] The lateral stability envelope is: ; in, The sideslip angle is the angle at the vehicle's center of gravity. This represents the upper boundary of the sideslip angle at the center of gravity; the formula means that the higher the vehicle speed and the more intense the driving intention, the smaller the allowable sideslip angle.

[0062] The tire utilization rate envelope is specifically as follows: Tire utilization envelope is used to constrain the overall tire utilization rate of each wheel: ; in, This represents the upper limit of tire utilization.

[0063] The upper bound of tire utilization is determined by the nominal threshold, the adhesion scaling factor, and the driving intention scaling factor: ; in, This represents the upper limit of nominal tire utilization.

[0064] The attachment scaling factor is: ; The driving intention scaling factor is: ; in, This is the attachment scaling factor. The reference road surface adhesion coefficient, and These are the lower and upper limits of the attachment scaling factor, respectively. This is the contraction coefficient of the tire utilization boundary based on driving intention.

[0065] The tire utilization envelope is: ; in, For the first The overall utilization rate of each wheel's tires This represents the upper boundary of tire utilization. The normal load transfer envelope includes normal load constraints for each round, lateral load transfer rate constraints, and longitudinal load transfer rate constraints, specifically: The normal load transfer envelope is used to limit excessively low or high normal loads on the wheels, as well as excessive longitudinal or lateral load transfer. The normal load constraints for each wheel are as follows: ; in, ; ; in, This is the proportional coefficient for the lower limit of the normal load. This is the upper limit proportional coefficient for the normal load.

[0066] Define the lateral load transfer rate: ; Define the longitudinal load transfer rate: ; And set constraints: ; ; in, This represents the upper limit of the lateral load transfer rate. This represents the upper limit of the longitudinal load transfer rate.

[0067] The actuator margin envelope includes steering system constraints, drive system constraints, braking system constraints, and active suspension system constraints, specifically: The actuator margin envelope is used to limit the amplitude and rate of change of the steering, drive, braking, and active suspension actuators. The steering system constraints are: ; The drive system constraints are: ; The braking system constraints are: ; The constraints of the active suspension system are: .

[0068] This implementation method constructs a four-dimensional dynamic safety envelope consisting of attitude stability dimension, tire adhesion utilization dimension, load transfer dimension, and actuator margin dimension to achieve a comprehensive characterization of the vehicle's safety margin under extreme operating conditions.

[0069] Furthermore, in one embodiment of this implementation, the four-dimensional dynamic safety envelope is corrected using a control invariant set, and the corrected four-dimensional dynamic safety envelope is scaled to obtain the final four-dimensional dynamic safety envelope.

[0070] The initial envelope obtained from the mechanism model does not fully consider the effects of modeling errors, actuator delays, tire nonlinearity, and disturbances. This implementation further uses a control invariant set to correct the initial envelope.

[0071] Define the nominal security set as: ; Based on discrete vehicle dynamics model: ; Offline construction of control invariant sets: ; in, For the first The candidate safe set for the next iteration To allow control over the set of inputs.

[0072] Once the set iterations converge, the corrected safety envelope is obtained: ; in, A modified safety envelope is provided to take controllability into account.

[0073] This means that not all theoretically safe states are truly controllable; only those states that can remain safe within the capabilities of the actuator are retained.

[0074] In another alternative implementation, a security envelope correction method based on predicted reachability domains can also be used.

[0075] Specifically, given the vehicle state, road surface adhesion coefficient, disturbance boundary, and actuator constraints, the set of states that the vehicle can reach within a finite time domain is predicted, and states that may cause yaw, side slip, tire utilization, load transfer, or actuator margin overrun are removed from the initial safety envelope; this reachability correction method can be used as an alternative or supplementary implementation of the control invariant set correction.

[0076] Vehicle safety boundaries change significantly with vehicle speed, road surface adhesion coefficient, road curvature, driving intention, and actuator state. For example, on low-adhesion surfaces, the permissible yaw rate and sideslip angle boundaries should be significantly narrowed; at high speeds, the same sideslip angle may correspond to a higher risk of instability; during emergency obstacle avoidance, although the vehicle needs rapid lateral maneuvering, tire adhesion margin and actuator control margin should be reserved in advance. Existing methods often use fixed thresholds or offline-calibrated fixed stability domains, making it difficult to update safety boundaries in real time according to the vehicle's current state. If the boundaries are set too wide, the controller may intervene too late, increasing the risk of instability; if the boundaries are set too narrow, the controller may intervene too early, resulting in decreased trajectory tracking capability, overly conservative control actions, and affecting driving comfort.

[0077] Therefore, to address the aforementioned technical issues, during online operation, the modified safety envelope is scaled based on the adhesion coefficient, vehicle speed, and driving intention: ; in: ; ; ; ; ; The yaw envelope scaling factor is: ; The final yaw boundary is: ; The final lateral boundary is: The aforementioned online scaling factors include at least the adhesion scaling factor, speed scaling factor, driving intention scaling factor, and actuator margin scaling factor. When the adhesion coefficient decreases, the vehicle speed increases, the driving intention strengthens, or the actuator margin decreases, the safety boundary of the corresponding dimension shrinks according to a preset scaling relationship; when the adhesion conditions improve, the vehicle moves away from the extreme operating condition, or the actuator margin recovers, the safety boundary of the corresponding dimension gradually widens at a preset rate to avoid abrupt changes in the control quantity caused by sudden changes in the safety envelope.

[0078] ; in, This is the scaling factor for the lateral boundary.

[0079] Implementation Method Four: The difference between this implementation method and Implementation Method One is that, in step 4, the vehicle status is divided into regions based on a unified safety margin, specifically as follows: Set stability threshold and critical threshold ,when At that time, the vehicle was in a stable zone, when At that time, the vehicle was in the critical zone, when At that time, the vehicle was in an unstable zone, among which, To standardize safety margins.

[0080] This implementation method obtains a unified safety margin based on normalized safety margins for yaw, side slip, tire adhesion utilization, load transfer, and actuator margin. Based on this unified safety margin, the vehicle state is divided into regions, such as... Figure 4 The diagram shown is a schematic diagram of state division.

[0081] Define the yaw safety margin: ; Define the lateral safety margin: ; Define the tire safety margin: ; Define the safety margin for normal loads: ; Define actuator safety margin: ; in, For the first The actuator in the first The actual control quantity at each control moment. For the first The maximum permissible control quantity for each actuator.

[0082] The unified safety margin is defined as: ; The unified safety margin is obtained by normalizing and fusing the yaw safety margin, sideslip safety margin, tire utilization safety margin, normal load safety margin, and actuator safety margin. The fusion method can be weighted summation, weighted minimum value, or a combination of both; when the safety margin of any dimension decreases, the unified safety margin decreases simultaneously, so that the area determination can reflect risks that precede attitude instability, such as tire adhesion saturation, excessive load transfer, or actuator approaching saturation.

[0083] In this embodiment, the stability threshold and critical threshold The data was calibrated using offline simulation data and real-vehicle test data. Specifically, vehicle operation data was collected under different vehicle speeds, different adhesion coefficients, different road curvatures, and different steering intensities. The samples were labeled as stable samples, critical samples, and unstable samples. The threshold was determined by minimizing the misjudgment cost function. ; in, The threshold misjudgment cost function, This is the probability that a stable sample is misclassified as a critical sample. This represents the probability that a critical sample is misclassified as an unstable sample. This represents the probability that an unstable sample is misclassified as a stable sample. , , These are the weights corresponding to the probability of misjudgment.

[0084] When the vehicle's state changes near the stability boundary, sensor noise, adhesion coefficient estimation error, tire force estimation error, and actuator delay can all cause the region determination result to frequently switch between the stable region, the critical region, and the unstable region. If the control system directly switches the control strategy based on the hard threshold, it is easy to cause control mode jitter, which in turn causes abrupt changes in steering, driving, braking, and suspension control quantities, reducing control smoothness and system robustness.

[0085] To address the aforementioned technical issues, in one embodiment of this implementation, the vehicle state is divided into a stable region, a critical region, and an unstable region by combining hysteresis logic or a fuzzy state machine.

[0086] To avoid frequent switching of vehicle states near the boundary, a first hysteresis band is set between the stable region and the critical region, and a second hysteresis band is set between the critical region and the unstable region.

[0087] Set an entry threshold between the stable region and the critical region. and exit threshold ,and: ; Set an entry threshold between the critical region and the unstable region. and exit threshold ,and: ; When a vehicle returns from the critical region to the stable region, the following conditions must be met: ; When a vehicle enters the critical region from the stable region, the following conditions must be met: ; When a vehicle moves from the critical region to the unstable region, the following conditions must be met: ; When a vehicle returns from the unstable region to the critical region, the following conditions must be met:

[0088] The hysteresis band width is: ; in, The hysteresis band width, To standardize the estimation of noise standard deviation for safety margin, To standardize the maximum rate of change of safety margin.

[0089] In another embodiment, a fuzzy state machine is used for risk assessment, such as... Figure 5 As shown. The input variables of the fuzzy state machine include a uniform safety margin. Unified safety margin change rate Road surface adhesion coefficient Intensity of driving intention and actuator saturation .

[0090] The uniform safety margin change rate is: ; The actuator saturation is: ; Fuzzy state machine output risk index: ; The risk index is obtained by defuzzifying the weighted average: ; in, To determine the number of fuzzy rules, For the first The fuzzy rule in the first Trigger intensity at each control moment, For the first The risk output value corresponding to each fuzzy rule.

[0091] Set the first risk threshold. Second risk threshold ,satisfy: ; when At that time, the vehicle was in a stable zone; when At that time, the vehicle was in the critical zone; when At that time, the vehicle was in an unstable zone.

[0092] To suppress fluctuations at the risk index boundary, a small hysteresis band is set: ; The width of the small hysteresis band is: ; in, The hysteresis band width of the risk index. The noise standard deviation of the risk index. This represents the maximum rate of change in the risk index.

[0093] This implementation differs from vehicle stability determination methods that only divide the stability domain in a two-dimensional phase plane of yaw rate and center of gravity sideslip angle. Two-dimensional phase plane methods typically only reflect the vehicle's planar attitude stability, while this implementation incorporates tire utilization, normal load transfer, and actuator margin into the dynamic safety envelope. This allows the control system to identify and intervene in advance when the yaw rate or center of gravity sideslip angle has not yet significantly exceeded its limits, but tire adhesion or actuator capability is already close to its limit.

[0094] Implementation Method 5: The difference between this implementation method and Implementation Method 1 is that, in step 5, the control weights of the control target are continuously adjusted based on the vehicle's location and the region's membership degree. Specifically: When the membership degree of the stable region increases, the weights of the trajectory tracking target and the actuator smoothness target are increased; When the membership degree of the critical region increases, the weights of the yaw stability target, the side slip stability target, the load transfer target, and the tire utilization control target are increased. When the membership degree of the unstable region increases, the weight of the safe recovery target is increased, and the weight of the trajectory tracking target is decreased.

[0095] Existing model predictive control methods typically superimpose trajectory error, yaw error, sideslip error, tire utilization, and control energy into a single objective function with fixed weights. Fixed weights fail to reflect changes in the priority of control objectives across different risk regions, easily leading to conflicts between stability and trajectory tracking objectives. Furthermore, the fixed priority of control objectives makes it difficult to adapt to varying risk states.

[0096] To address the aforementioned issues, this implementation proposes a multi-actuator collaborative chassis stability control method capable of dynamically adjusting the priority of control objectives based on the vehicle's risk zone. Different control objective priorities are generated according to the vehicle's location. These objectives include trajectory tracking, yaw stability, sideslip stability, tire utilization control, load transfer control, actuator ride comfort, and safety recovery. When the vehicle is in a stable zone, the control objectives should primarily focus on trajectory tracking, control ride comfort, and overall comfort. When the vehicle is in a critical zone, the weights of yaw stability, sideslip stability, tire utilization constraints, and load transfer constraints should be increased. When the vehicle is in an unstable zone, returning to the vehicle's safe envelope should be the highest priority, allowing for a short-term sacrifice of some trajectory tracking performance. Specifically: Step 51, Stable region control target When the vehicle is in the stable region, its state is far from the safety envelope boundary, and the control objective is primarily trajectory tracking and control smoothness. The objective function for the stable region is: ; in, For the objective function in the stable region, To predict the number of steps in the time domain, For the horizontal error weight in the stable region, As the heading error weight in the stable region, Incremental weights are controlled for the stable region.

[0097] Step 52, Critical Zone Control Target When the vehicle is in the critical region, its state approaches the dynamic safety envelope boundary. The control objectives are primarily yaw stability, sideslip stability, tire utilization constraints, and load transfer constraints, while also considering trajectory tracking and control ride comfort. The objective function for the critical region is: ; in, Let the objective function be the one for the critical region. For reference yaw rate, For the critical region oscillation stability weight, As the critical region side-bias stability weight, As the weight for tire utilization rate in the critical zone, This represents the expected tire utilization rate. and These are the transverse load transfer weights and longitudinal load transfer weights in the critical region, respectively. and These are the warning thresholds for lateral load transfer rate and longitudinal load transfer rate, respectively. The weighting of the transverse trajectory error in the critical region. Incremental weights are used to control the critical region.

[0098] The load transfer rate warning threshold is less than the corresponding dynamic safety envelope upper limit, that is:

[0099] This allows the load transfer penalty to intervene before the vehicle reaches the boundary of the normal load transfer envelope.

[0100] Step 53, Unstable Region Control Target When a vehicle is in an unstable region, its state has approached or exceeded the safety boundary, and the control objective is primarily to restore safety. The normalized distance by which the vehicle state exceeds the dynamic safety envelope is defined as follows: ; The objective function for the unstable region is: ; in, For the objective function in the unstable region, To safely restore the weights of the envelope, To suppress the weight of horizontal movement, For lateral bias suppression weights, The weight for suppressing tire utilization rate is determined.

[0101] Step 54, control the continuous transition of weights To avoid abrupt changes in control objectives due to region switching, the weights of the control objectives are continuously adjusted based on region membership. Specifically, Increase the weights for trajectory tracking and actuator smoothness when the membership degree of the stable region increases; Increase the membership degree of the critical region to improve the weights of yaw stability, sideslip stability, tire utilization, and load transfer constraints. When the membership degree in the unstable region increases, the weights for safe recovery and boundary violation suppression are increased, while the weights for trajectory tracking are allowed to decrease temporarily. ; in, For the first The control objective is in the first The actual weight at each control moment , , Let these be the regional membership degrees of the stable region, critical region, and unstable region, respectively, and satisfy the following: .

[0102] This implementation differs from the conventional yaw-roll model predictive control method. Conventional yaw-roll MPC (model predictive control) typically optimizes yaw and roll responses simultaneously with fixed weights, and uses the active suspension primarily as a roll suppression actuator. This invention uses a four-dimensional dynamic safety envelope as the basis for state constraints and target priority generation, and continuously changes the control weights according to the regional membership of the stable, critical, and unstable regions, so that the priorities among trajectory tracking, yaw stabilization, side slip stabilization, tire utilization, load transfer, and actuator ride comfort adaptively change with the risk state.

[0103] Implementation Method Six: The difference between this implementation method and Implementation Method One is that step 6 uses a hierarchical optimization method to solve for the multi-actuator cooperative control quantity, specifically: The upper layer is based on the vehicle dynamics model and the four-dimensional dynamic safety envelope, with the control weight as the weighting coefficient of the objective function. The nonlinear model predicts and controls the front wheel steering angle, the desired longitudinal resultant force, the desired additional yaw moment, and the desired roll control moment. The lower layer distributes the desired longitudinal resultant force and the desired additional yaw moment into torque commands for each wheel drive motor and braking torque commands for each wheel, and distributes the desired roll control moment into power commands for each wheel active suspension.

[0104] Distributed drive electric vehicles offer a high degree of control freedom because each wheel can independently output driving or braking torque; steer-by-wire systems can actively adjust the front wheel angle; and active suspension systems can adjust vehicle posture and tire normal load distribution. Therefore, the combined participation of actuators such as steering, driving, braking, and active suspension in chassis control has become a key technological direction for improving vehicle handling stability, trajectory tracking performance, and safety under extreme conditions.

[0105] Existing stability control methods mostly focus on one type of actuator in differential braking, differential drive, or active steering. For example, differential braking generates additional yaw moment, wheel-side drive torque distribution improves yaw response, or active steering corrects the vehicle path. However, under extreme conditions, a single actuator is prone to insufficient control capability or over-intervention. Active suspension can improve tire adhesion utilization by adjusting body roll and tire normal load, but in existing chassis control methods, active suspension is usually used as a comfort control actuator and rarely works in conjunction with steering, drive, and braking actuators within the same stability control framework. Therefore, existing methods struggle to fully utilize the complementary advantages of multiple actuators.

[0106] To address the aforementioned issues, this embodiment proposes a multi-actuator collaborative chassis stability control method that performs hierarchical optimization allocation among steering, driving, braking, and active suspension. Based on the vehicle's current state, four-dimensional dynamic safety envelope, the vehicle's location within a risk zone, and the priority of control objectives, the multi-actuator control quantities used for vehicle stability control are solved. To avoid excessive optimization dimensionality, excessive computational burden, and duplication of lower-level actuator allocation logic due to the upper-level controller directly solving for all actuator control quantities, this embodiment employs a hierarchical optimization method that solves for generalized control requirements at the upper level and allocates specific actuators at the lower level. Figure 6 As shown.

[0107] Specifically, the upper-level controller, based on the vehicle dynamics model and the four-dimensional dynamic safety envelope, uses nonlinear model predictive control to solve for the front wheel steering angle, desired longitudinal resultant force, additional yaw moment, and roll control moment; the lower-level controller, based on the desired longitudinal resultant force, additional yaw moment, and roll control moment, as well as tire adhesion constraints, actuator amplitude constraints, and actuator rate of change constraints, allocates them as four-wheel drive motor torque commands, four-wheel braking torque commands, and four-wheel active suspension power commands.

[0108] In this embodiment, the upper-level controller does not directly output the torque of each wheel drive motor. Braking torque of each wheel and each active suspension as power Instead, it outputs a generalized control quantity that characterizes the vehicle's stability control requirements. The specific actuator control quantity is obtained by solving the lower-level actuator allocation circuit.

[0109] Step 61: Establish the upper-level generalized control prediction model Define the upper-level controller state vector as follows: ; in, For the first The upper-level controller state vector at each control moment; The longitudinal speed of the vehicle; The vehicle's lateral speed; The vehicle's yaw rate; The sideslip angle is the angle between the vehicle's center of gravity and its body. This refers to the vehicle body roll angle; The vehicle body roll rate; This represents the lateral position error of the vehicle relative to the reference trajectory. This represents the heading error between the vehicle's heading angle and the tangent direction of the reference trajectory.

[0110] Define the upper-level generalized control input vector as: ; in, For the first The upper-level generalized control input vector at each control moment; The steering angle of the front wheels; To achieve the desired longitudinal resultant force; To add a desired yaw moment; The desired roll control moment.

[0111] The upper-level vehicle dynamics prediction model is represented as follows: ; in, This is a discrete vehicle dynamics prediction model for the upper layer. For the first The road surface adhesion coefficient estimated at each control time.

[0112] The upper-level vehicle dynamics prediction model includes vehicle longitudinal dynamics, lateral dynamics, yaw dynamics, center of gravity sideslip angle dynamics, vehicle roll dynamics, and trajectory error dynamics.

[0113] The longitudinal dynamics of the vehicle are expressed as: ; in, For the overall vehicle weight; This refers to the resistance force on the vehicle.

[0114] The lateral dynamics of a vehicle are expressed as: ; in, The equivalent lateral tire force is determined by the front wheel steering angle, vehicle condition, road surface adhesion coefficient, and tire model.

[0115] In one embodiment, the equivalent lateral tire force can be expressed as: ; in, This is the equivalent lateral tire force on the front axle. This is the equivalent lateral tire force on the rear axle.

[0116] The vehicle yaw dynamics are expressed as: ; in, Let yaw moment of inertia be the moment of inertia of the vehicle about its vertical axis. The basic yaw moment is generated by the front wheel steering angle and the lateral force of the tires; The desired additional yaw moment generated by the differential distribution of drive and braking.

[0117] In one embodiment, the basic yaw moment can be expressed as: ; in, This is the distance from the vehicle's center of gravity to the front axle. This is the distance from the vehicle's center of gravity to the rear axle.

[0118] The dynamic expression of the center of mass sideslip angle is: ; in, To prevent positive numbers with a denominator of zero.

[0119] Vehicle body roll dynamics are expressed as: ; in, The moment of inertia of the vehicle body about its longitudinal axis is the roll inertia. This is the equivalent roll damping; This is the equivalent roll stiffness; For the sprung mass; The height from the center of mass of the sprung mass to the center of tilt. The desired roll control moment.

[0120] The trajectory error dynamics are expressed as: ; ; in, The curvature is used as a reference trajectory.

[0121] Normal load of each wheel In the prediction time domain, these are treated as algebraic variables, not as independent integration states. Specifically, the normal loads on each wheel are updated in real time based on longitudinal acceleration, lateral acceleration, vehicle roll state, and desired roll control moment through load transfer relationships. ; in, In the first The first control time prediction Step 1 Normal load on each wheel; For the first Static normal load on each wheel; For longitudinal load transfer terms; The table shows the lateral load transfer terms; This is the active load adjustment term caused by the equivalent desired roll control moment.

[0122] Step 62: Establish the upper-level nonlinear model predictive control problem driven by the region. Within each control cycle, based on the current estimated state Estimated value of road surface adhesion coefficient Four-dimensional dynamic security envelope A nonlinear predictive control problem is established based on the region where the vehicle is located. The state recursion relationship in the prediction time domain is as follows: ; in, ; To predict the length of the time domain; In the first The first control time prediction Step state; In the first The first control time prediction Step generalized control input.

[0123] The initial conditions are: ; in, For the first State estimates at each control moment.

[0124] Select the corresponding objective function based on the region where the vehicle is located: ; in, The objective function in the stable region focuses on trajectory tracking and control smoothness. The objective function for the critical region focuses on yaw stability, lateral stability, and tire utilization control. The objective function for the unstable region focuses on the four-dimensional dynamic safety envelope of vehicle state regression.

[0125] In one implementation, the objective function for the stable region is expressed as: ; The objective function for the critical section is expressed as: ; The objective function for the unstable region is expressed as: ; in, For reference yaw rate; This represents the upper limit of the expected tire utilization rate. To predict the normalized out-of-bounds distance of the state relative to the four-dimensional dynamic safety envelope; , , , , , , , , , , , and These are the weighting coefficients for the corresponding control objectives.

[0126] The normalized out-of-bounds distance of the predicted state relative to the four-dimensional dynamic safety envelope can be expressed as: ; The upper-level nonlinear model predictive control optimization problem is expressed as: ; in, To predict the generalized control input sequence in the time domain: ; in, To control the length of the time domain; Slack variables for safety constraints; This is a penalty term for slack variables in safety constraints.

[0127] In one implementation, the penalty term for the slack variable of the safety constraint is expressed as: ; in, Slack variable weights for safety constraints.

[0128] The constraints of the upper-level optimization problem include: ; ; ; ; in, For the set of generalized control input amplitude constraints; For the set of generalized control input rate of change constraints; For the relaxation extension of the safety envelope constraint.

[0129] After solving the above optimization problem, the optimal generalized control input sequence is obtained: ; In rolling optimization control, only the first term of the optimal control sequence is used as the upper-level output of the current control cycle: ; in, The desired front wheel steering angle obtained from the upper-level optimization; The desired longitudinal resultant force obtained from the upper-level optimization; Add a yaw moment to the desired result obtained from the upper-level optimization; The desired roll control torque is obtained from the upper-level optimization.

[0130] Step 63, Lower-level drive and brake actuator allocation The lower-level drive and brake actuators are allocated to distribute the desired longitudinal resultant force. and expected additional yaw moment The force is distributed to the longitudinal tires of the four wheels and further converted into torque commands and braking torque commands for each wheel drive motor.

[0131] Define the longitudinal tire force vector of the four wheels as: ; The mapping relationship between the longitudinal tire forces of the four wheels and the resultant longitudinal force and additional yaw moment of the vehicle is as follows: ; in, The resultant force of the longitudinal forces of the four wheels. This is the additional yaw moment generated by the difference in longitudinal force between the left and right wheels; This is the longitudinal force distribution matrix. For the left and right wheelbases... For vehicles, if the vehicle coordinate system specifies that the left side is positive and the right side is negative, then It can be represented as: ; It should be noted that the above The direction of the symbols in the second row can be adjusted equivalently according to the vehicle coordinate system and the definition of the positive direction of the yaw moment.

[0132] The drive and braking distribution optimization problem is expressed as: ; in, ; in, The weighting matrix for tracking longitudinal resultant force and yaw moment; This is the penalty coefficient for longitudinal tire force variation; This is the penalty coefficient for tire overall utilization rate; This refers to the longitudinal tire force of the four wheels in the previous control cycle.

[0133] Drive and braking distribution constraints include: ; ; ; in, and The first The lower and upper limits of longitudinal tire force for each wheel; For the first Maximum change in longitudinal tire force of each wheel; For the first Lateral tire force of each wheel; For the first Normal load on each wheel.

[0134] The optimal longitudinal tire force is obtained by solving the problem. Then, based on its sign, it is converted into a drive motor torque command or a braking torque command. When: ; When, it indicates the first Each wheel is in a driving demand state, and its command is: ; ; when: ; When, it indicates the first When a wheel is in braking demand, the instruction is: ; ; in, For the first Torque command for the drive motor of each wheel. For the first Braking torque command for each wheel. This refers to the effective rolling radius of the tire. By using the above conversion rules, simultaneous driving and braking of the same wheel can be avoided, improving the engineering feasibility of control distribution.

[0135] Step 64, Lower-level active suspension actuator allocation The lower active suspension actuator is used to distribute the desired roll control torque. The power commands are assigned to four active suspensions.

[0136] Define the active suspension as the force vector: ; The mapping relationship between the active suspension's dynamic force and the roll control torque is as follows: ; in, This refers to the actual roll control moment generated by the active suspension. Create a power distribution matrix for the active suspension. For left and right track widths... For a vehicle, if the vehicle coordinate system and the positive direction of the roll moment are defined with the left side being positive and the right side being negative, then... It can be represented as: ; It should be noted that the above The directions of the various symbols can be adjusted equivalently according to the vehicle coordinate system and the definition of the positive direction of the roll moment.

[0137] The active suspension allocation optimization problem is expressed as: ; in, ; The weighting is used to track the roll control torque. Apply a penalty coefficient for dynamic changes to the active suspension. Assign a penalty factor to the normal load; It provides power to the active suspension of the previous control cycle; For the first The expected normal load of each wheel.

[0138] Active suspension distribution constraints include: ; ; ; in, For the first The maximum operating force of the active suspension; For the first The maximum change in the dynamic force of the active suspension; For the first The maximum permissible relative displacement of the suspension.

[0139] After obtaining the optimal active suspension action force, the active suspension action force command is generated: ; in, For the first One active suspension system provides power commands; To optimize the obtained first An active suspension system provides power.

[0140] Step 65, front wheel steering angle command generation The desired front wheel steering angle is obtained by upper-level nonlinear model predictive control. Then, the steering allocation unit generates front wheel steering angle commands based on the front wheel steering angle amplitude constraint and the steering angle change rate constraint.

[0141] The front wheel steering angle amplitude constraint is: ; The constraint on the rate of change of front wheel steering angle is: ; in, For the first Front wheel steering angle command at each control moment; This is the maximum steering angle of the front wheels; This indicates the maximum change in the front wheel steering angle within a single control cycle.

[0142] In one implementation, the front wheel steering angle command is expressed as: ; in, This is the amplitude limiting function.

[0143] Step 66, multi-actuator control output After the above-mentioned hierarchical optimization, the control allocation module outputs the final multi-actuator control instructions: ; in, For the first The control command vector is ultimately output to the vehicle actuator group at each control moment.

[0144] Through the hierarchical optimization method described above, the upper-level controller is only responsible for determining the front wheel steering angle and generalized control requirements needed for vehicle stability control at the vehicle level. The lower-level controller then allocates the generalized control requirements to specific actuators based on the capabilities of each actuator, tire adhesion constraints, and actuator rate of change constraints. This method avoids the problems of excessive optimization dimensionality and repetitive control logic caused by the upper-level controller directly solving for all actuator control quantities. At the same time, it can fully leverage the coordinated control capabilities between front-wheel steering, four-wheel drive, four-wheel braking, and active suspension.

[0145] This implementation also treats actuator margin as part of the safety envelope, rather than just as a lower-level limiting condition. Therefore, when the steering, drive, braking, or active suspension actuators approach amplitude or rate of change limits, the upper-level safety envelope and control target can be simultaneously contracted or reconfigured, preventing upper-level control requirements from exceeding the capabilities of lower-level actuators, thereby improving the feasibility and robustness of control allocation under extreme conditions.

[0146] In summary, this embodiment discloses a multi-actuator collaborative chassis stability control method, system, vehicle, and storage medium based on a four-dimensional dynamic safety envelope, belonging to the field of vehicle dynamics control and intelligent chassis control technology. The method includes: collecting vehicle operating state information, actuator feedback information, and road environment information; estimating vehicle state, road surface adhesion, tire utilization rate, normal load, and actuator remaining margin online; constructing a four-dimensional dynamic safety envelope composed of attitude stability dimension, tire adhesion utilization dimension, load transfer dimension, and actuator margin dimension; obtaining a unified safety margin based on the normalized safety margins of each dimension, and dividing the vehicle state into stable, critical, and unstable regions using hysteresis logic or a fuzzy state machine; continuously adjusting the weights of trajectory tracking, yaw stability, sideslip stability, tire utilization rate, load transfer, and actuator ride comfort according to the region membership degree; and employing a hierarchical optimization method to solve for the front wheel steering angle, desired longitudinal resultant force, additional yaw moment, and roll control moment at the upper level, and then allocating them to four-wheel drive / braking torque and active suspension for power at the lower level. This implementation can improve the stability, trajectory tracking performance and control robustness of vehicles under complex conditions such as low adhesion, high-speed obstacle avoidance and extreme steering.

[0147] Implementation Method Seven: The difference between this implementation method and Implementation Method One is that the method further includes: Step 7: Obtain actuator feedback and vehicle response, update vehicle state estimation, four-dimensional dynamic safety envelope and control target based on actuator feedback and vehicle response, and enter the next control cycle.

[0148] This embodiment is used in the first After the multi-actuator control command is executed in each control cycle, the vehicle state estimation result, four-dimensional dynamic safety envelope, vehicle risk area and control target priority are updated based on the actual vehicle response and actuator feedback information, thereby forming a rolling closed-loop control.

[0149] Control distribution module 8 sends front wheel steering angle control commands Drive torque command Braking torque command and active suspension as power command The signals are respectively sent to the front-wheel steering system 91, the four-wheel drive system 92, the four-wheel braking system 93, and the active suspension system 94 of vehicle 1, such as... Figure 7 As shown.

[0150] After actuator group 9 completes its action, it feeds back the actual execution amount to closed-loop feedback update module 10. The actual execution amount includes the actual front wheel steering angle. Actual driving torque Actual braking torque and actual active suspension as power The closed-loop feedback update module 10 re-inputs the actual execution quantities, inertial measurement unit measurements, wheel speed measurements, and suspension measurements into the parameter estimation module 3, updating the vehicle's longitudinal speed, lateral speed, center of gravity sideslip angle, road adhesion coefficient, normal load on each wheel, and overall tire utilization rate. ; The four-dimensional dynamic safety envelope construction module 4 updates the yaw stability envelope, side slip stability envelope, tire utilization envelope, normal load transfer envelope, and actuator margin envelope based on the updated road adhesion coefficient, vehicle longitudinal speed, driving intention intensity, estimated normal load values ​​for each wheel, and actuator remaining margin. The four-dimensional dynamic safety envelope at each control moment is represented as: ; Region determination module 5 recalculates the unified security margin based on the updated four-dimensional dynamic security envelope. or risk index The system updates the vehicle region status based on hysteresis logic. If the vehicle returns from the unstable region to the critical or stable region, the weight of the safety recovery target is gradually reduced, and the trajectory tracking target and ride comfort target are restored. If the vehicle enters the critical region from the stable region, the weights of the yaw stability, side slip stability, and tire utilization control targets are increased in advance. If the vehicle enters the unstable region, the weight of the safety recovery target is immediately increased, and the trajectory tracking target is allowed to be temporarily de-weighted.

[0151] After completing the above updates, the number will be... The state estimation results at each control moment, the four-dimensional dynamic safety envelope, the vehicle region state, and the control target weights are used as inputs for the next control cycle. The steps of state acquisition, parameter estimation, four-dimensional dynamic safety envelope construction, region determination, target priority generation, and hierarchical optimization allocation are re-executed, thereby realizing the closed-loop rolling update of vehicle chassis stability control.

[0152] Furthermore, this can serve as a further technical feature for subsequent subordinate protection: First, the four-dimensional dynamic safety envelope can be further defined by using control invariant set correction, and in alternative implementations, predictive reachability domain correction can be used to enhance the adaptability of the safety envelope to actuator constraints, modeling errors and disturbance boundaries.

[0153] Second, the online scaling factor can be further defined as being determined by at least one of the following: road surface adhesion coefficient, vehicle speed, driving intention intensity, road curvature, obstacle risk, and actuator margin, and applied to the yaw boundary, side slip boundary, tire utilization boundary, load transfer boundary, and actuator margin boundary, respectively.

[0154] Third, the unified safety margin formula can be further defined as normalizing and integrating the yaw safety margin, side slip safety margin, tire utilization safety margin, normal load safety margin, and actuator safety margin, and dividing the stable region, critical region, and unstable region according to the integration result.

[0155] Fourth, the hysteresis band width can be further limited to be determined based on the noise standard deviation estimated by the uniform safety margin and the maximum rate of change of the uniform safety margin, thereby suppressing frequent switching of vehicle states near the region boundary.

[0156] Fifth, the fuzzy state machine can be further defined as taking a unified safety margin, a unified safety margin change rate, a road surface adhesion coefficient, a driving intention intensity, and an actuator saturation as inputs, and a risk index as output, and determining the vehicle risk area through the risk index threshold and the small hysteresis band.

[0157] Sixth, the drive / brake actuator allocation can be further limited to mutual exclusion conversion of drive and brake based on the sign of longitudinal tire force, so that the same wheel does not output drive torque and brake torque at the same time in the same control cycle.

[0158] Seventh, the active suspension actuator allocation can be further defined as optimizing the four active suspension actuators based on the desired roll control torque and desired normal load distribution, in order to adjust the normal load on the left and right or front and rear wheels and reduce the peak of the overall tire utilization rate.

[0159] Eighth, this embodiment can also be extended to a corresponding multi-actuator collaborative chassis stability control system, vehicle, and computer-readable storage medium. The system includes a parameter estimation module, a four-dimensional dynamic safety envelope construction module, a region determination module, a control target priority generation module, an upper-level optimization module, a lower-level control allocation module, and a closed-loop feedback update module.

[0160] The above provides a detailed description of the multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope proposed in this invention. Specific examples have been used to illustrate the principles and implementation methods of this invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this invention. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this invention. Therefore, the content of this specification should not be construed as a limitation of this invention.

Claims

1. A multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope, characterized in that, Includes the following steps: Step 1: Collect vehicle operating status information, actuator feedback information, and road environment information; Step 2: Estimate the vehicle state based on vehicle operating status information, actuator feedback information and road environment information, and construct a four-dimensional dynamic safety envelope based on the estimation results; Step 3: Calculate the safety margin corresponding to each envelope in the four-dimensional dynamic safety envelope, and normalize and merge the safety margins corresponding to each envelope into a unified safety margin. Step 4: Based on a unified safety margin, divide the vehicle status into regions, and obtain the region where the vehicle is located and the region membership degree based on the division results. Step 5: Continuously adjust the control weights of the control target based on the vehicle's location and the region's membership degree; Step 6: Based on the control weights, use a hierarchical optimization method to solve the multi-actuator collaborative control quantity, and output the multi-actuator collaborative control quantity to the corresponding actuator of the vehicle to complete the chassis stability control of the current control cycle.

2. The multi-actuator collaborative chassis stability control method based on four-dimensional dynamic safety envelope according to claim 1, characterized in that, The vehicle state estimation in step 2 includes the vehicle's longitudinal speed, lateral speed, yaw rate, center of gravity sideslip angle, road surface adhesion coefficient, tire utilization rate, normal load of each wheel, and remaining margin of steering, driving, braking, and active suspension actuators.

3. The method of claim 1, wherein, The four-dimensional dynamic security envelope in step 2 is specifically as follows: ； wherein, is a yaw stability envelope, is a side slip stability envelope, is a tire utilization envelope, is a normal load transfer envelope, is an actuator margin envelope.

4. The method of claim 3, wherein, The yaw stabilizing envelope is specifically as follows: ； in, The initial upper boundary of the yaw rate. This is the vehicle state vector. The vehicle's yaw rate; The lateral stabilizing envelope is specifically: ； in, The sideslip angle is the angle at the vehicle's center of gravity. The upper boundary of the centroid sideslip angle; The tire utilization rate envelope is specifically as follows: ； in, For the first The overall utilization rate of each wheel's tires This represents the upper boundary of tire utilization. The normal load transfer envelope includes normal load constraints for each round, lateral load transfer rate constraints, and longitudinal load transfer rate constraints. The actuator margin envelope includes steering system constraints, drive system constraints, braking system constraints, and active suspension system constraints.

5. The multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope according to claim 1 or 3, characterized in that, The four-dimensional dynamic safety envelope is corrected by using a control invariant set, and then scaled to obtain the final four-dimensional dynamic safety envelope.

6. The multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope according to claim 1, characterized in that, In step 4, the vehicle status is divided into regions based on a unified safety margin, specifically as follows: Set stability threshold and critical threshold ,when At that time, the vehicle was in a stable zone, when At that time, the vehicle was in the critical zone, when At that time, the vehicle was in an unstable zone, among which, To standardize safety margins.

7. The multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope according to claim 1 or 6, characterized in that, The region division is determined using hysteresis logic or fuzzy state machine.

8. The multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope according to claim 1, characterized in that, The step 5, which involves continuously adjusting the control weights of the control target based on the vehicle's location and its regional membership, specifically involves: When the membership degree of the stable region increases, the weights of the trajectory tracking target and the actuator smoothness target are increased; When the membership degree of the critical region increases, the weights of the yaw stability target, the side slip stability target, the load transfer target, and the tire utilization control target are increased. When the membership degree of the unstable region increases, the weight of the safe recovery target is increased, and the weight of the trajectory tracking target is decreased.

9. The multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope according to claim 1, characterized in that, Step 6, which uses a hierarchical optimization method to solve for the multi-actuator cooperative control quantity, specifically involves: The upper layer is based on the vehicle dynamics model and the four-dimensional dynamic safety envelope, with the control weight as the weighting coefficient of the objective function. The nonlinear model predicts and controls the front wheel steering angle, the desired longitudinal resultant force, the desired additional yaw moment, and the desired roll control moment. The lower layer distributes the desired longitudinal resultant force and the desired additional yaw moment into torque commands for each wheel drive motor and braking torque commands for each wheel, and distributes the desired roll control moment into power commands for each wheel active suspension.

10. The multi-actuator cooperative chassis stability control method based on four-dimensional dynamic safety envelope according to claim 1, characterized in that, The method further includes: Step 7: Obtain actuator feedback and vehicle response, update vehicle state estimation, four-dimensional dynamic safety envelope and control target based on actuator feedback and vehicle response, and enter the next control cycle.

Images