Search

CN-121989915-A - Distributed driving vehicle stability control method, system and medium

CN121989915ACN 121989915 ACN121989915 ACN 121989915ACN-121989915-A

Abstract

The invention relates to the technical field of intelligent control of vehicles, in particular to a distributed driving vehicle stability control method, system and medium. The method comprises the following steps of S1, collecting real-time state data of a vehicle, calculating a transient load transfer index, calculating time lag compensation time of an actuator based on a secondary transient instability early warning algorithm to judge whether to trigger control intervention, S2, constructing a nominal dynamics model containing time lag of the actuator, learning residual errors between actual vehicle dynamics and the nominal model by using a sparse Gaussian process, constructing a data-driven time lag compensation model, S3, based on the time lag compensation model, adopting an anti-time lag optimal control strategy, adopting a random model to predict a control frame, constructing and solving a constraint optimization problem in a limited time domain, and realizing transient stability control of the vehicle based on optimal control input.

Inventors

  • LI GUOQIANG
  • MA HE

Assignees

  • 北京理工大学

Dates

Publication Date
20260508
Application Date
20260320

Claims (10)

  1. 1. A distributed drive vehicle stability control method, characterized by comprising the steps of: S1, collecting real-time state data of a vehicle, calculating a transient load transfer index, and calculating time lag compensation time of an actuator based on a secondary transient instability early warning algorithm so as to judge whether control intervention is triggered or not; S2, constructing a nominal dynamics model containing actuator time lag, and utilizing a sparse Gaussian process to learn residual errors between actual vehicle dynamics and the nominal model to construct a data-driven time lag compensation model; S3, based on a time lag compensation model, adopting an optimal control strategy for resisting time lag, adopting a random model prediction control framework, constructing and solving a constraint optimization problem in a limited time domain, and realizing transient stability control of the vehicle based on optimal control input.
  2. 2. The method for controlling stability of a distributed driving vehicle according to claim 1, wherein the collecting real-time state data of the vehicle, calculating a transient load transfer index specifically includes: Vehicle state data are collected in real time, and the transient load transfer index TLTI value at the current moment is calculated according to the following formula: (1), wherein T represents the track width of the vehicle, , , , , Of the above-mentioned parameters, For the corresponding unsprung mass of each wheel, For vertical acceleration of the unsprung mass of each wheel, The roll moment of inertia of the tractor and the semitrailer respectively, The roll acceleration of the tractor and the semitrailer respectively, In the form of a sprung mass, G is the total mass of the whole vehicle, g is the gravity acceleration, As the vertical acceleration of the sprung mass, The heights of the centers of gravity of the tractor and the semitrailer respectively, Lateral acceleration of the tractor and the semitrailer, respectively.
  3. 3. The method for controlling stability of a distributed driving vehicle according to claim 1, wherein the calculating the actuator time lag compensation time based on the secondary transient instability early warning algorithm specifically comprises: Defining a short-term prediction equation: (2), Calculating TLTI the first derivative and the second derivative by using a difference method, and calculating the residual time required for reaching a preset instability threshold from the current moment, namely the time lag compensation time ADCT of the actuator: (3), Wherein TLTI is the transient load transfer index.
  4. 4. The distributed drive vehicle stability control method according to claim 1, wherein the S1 further includes: when the calculated actuator time lag compensation time ADCT is less than the critical activation threshold When the system is in imminent instability, the trigger zone bit is immediately set =1, Activating the subsequent optimal control strategy, only when the actuator time lag compensation time ADCT rises back to exceed the activation threshold and the buffering time When the sum is over, the vehicle is judged to be safe to recover, and the trigger zone bit is set =0, Giving control right back to the driver, and the calculation method is as follows: (4), Wherein, the For the critical activation threshold value, To delay the buffering term, the robustness of the trigger mechanism to transient interference is enhanced, Representing the trigger state of the previous time step.
  5. 5. The method for controlling stability of a distributed driving vehicle according to claim 1, wherein the constructing a nominal dynamics model including actuator time lags in S2 is specifically: Quantitatively modeling the total time lag affecting the transient stability of the distributed driving semi-trailer, and defining the total time lag of a system total set as the linear superposition of CAN bus communication delay and the maximum inherent delay of an actuator, namely: (5), Wherein, the The time lag is summed up for the system total, For the CAN bus communication delay, For the actuator maximum inherent delay, the actuator maximum inherent delay is determined by a larger value of the steering system delay and the in-wheel motor driving system delay, that is: (6), considering the randomness of the CAN bus load, assuming uniform distribution over a communication delay obeying interval, the probability density function is as follows: (7), To adapt the discrete sampling characteristics of the controller, the continuous total time lag is converted into an equivalent discrete number of time steps d. Processing with a down-rounding function: (8), Wherein, the Sampling period of the controller; Based on this, a nominal kinetic model is constructed containing the input time lags: (9), Wherein, the Is a vehicle state vector, and is a control input after a delay of d steps.
  6. 6. The method for controlling stability of a distributed driving vehicle according to claim 1, wherein in S2, a residual error between actual vehicle dynamics and a nominal model is learned by using a sparse gaussian process, and constructing a data-driven time lag compensation model is specifically: Model errors are defined as the difference between the actual vehicle state and the nominal model state, namely: (10), Wherein, the For the model prediction residual error, In order to be in the actual state of the vehicle, Is a nominal model state; Collecting running data of vehicles at different speeds, loads and road adhesion coefficients, and constructing a training data set Wherein the regression vector Consisting of vehicle state and control inputs at the current moment, i.e. Target output Corresponding to the observed error value; Assuming that the errors of the state dimensions are independent of each other, for each output dimension Establishing an independent GP model, and measuring the correlation between samples by adopting a square index kernel function: (11), Wherein, the For the signal variance to be a function of the signal variance, Is a characteristic length scale matrix; the SGP model parameters obtained through offline training are deployed into a vehicle-mounted controller, online compensation is executed, and in each control step k, the SGP model parameters are based on the current regression vector Calculating posterior prediction mean and variance of model mismatch: (12), Wherein, the To induce a covariance matrix between the points, Covariance vector between current input and induction point; combining the predicted mean values of the dimensions into an error vector And injecting the final corrected state estimation into a nominal model to obtain the final corrected state estimation: (13), Wherein, the Representing the estimated state vector of the object, To map the error vector to a distribution matrix of the state space.
  7. 7. The method for controlling the stability of a distributed driving vehicle according to claim 1, wherein in the step S3, based on a time lag compensation model, an optimal control strategy for resisting time lag is adopted, a random model prediction control framework is adopted, and the constraint optimization problem in a limited domain is specifically established as follows: At each control moment Control input at one time above the controller And the current sensor-collected vehicle state For initial conditions, the modified state estimate is used as a predictive equation: (14), Wherein, the A state vector representing the predicted future k +1 step at the moment, To include a nominal model of the input skew, Model mismatch mean values predicted for the SGP model based on future states; in order to achieve the balance of the path tracking precision, the running stability and the smoothness of the control action of the vehicle, the following quadratic programming optimization problem is constructed: (15), Wherein, the To reflect the deviation between the actual driving track of the vehicle and the ideal reference track, is a state weight matrix, To adjust the control priority for the different state variables, In order to control the amplitude of the input, For inputting the weight matrix, the method is used for limiting the oversized control action, reducing the energy consumption, In order to control the rate of change of the quantity, The incremental weight matrix is used for preventing high-frequency jitter and ensuring the smooth operation of the actuator.
  8. 8. The method for controlling the stability of a distributed driving vehicle according to claim 1, wherein the solving the constraint optimization problem in the finite time domain in S3, and the controlling the transient stability of the vehicle based on the optimal control input is specifically: control input vector to be solved Comprising three key dimensions: (16), Wherein is The front wheel of the tractor is actively steered at an angle, For the additional yaw moment applied to the tractor, For an additional yaw moment applied to the semi-trailer; Using opportunistic constraints to handle state constraints, the probability of setting a vehicle state to remain within a safe area is greater than a set confidence level : (17), Converting the probability constraint into deterministic linear inequality constraint based on the prediction variance output by SGP model, and constraining boundary for each state Tightening it into: (18), Wherein, the For the state prediction mean value, For the state covariance resulting from SGP variance propagation, An inverse cumulative distribution function of the standard normal distribution; physical limitations are imposed on the control inputs and their increments: (19), Solving the optimization problem converted into deterministic constraint in each control period to obtain the future Optimal control delta sequence of steps According to the principle of rolling time domain control, only the first element in the sequence is selected Acting on the system to calculate the current actual control command And break it down into specific actuator instructions that will To the steer-by-wire system And Converting the torque distribution algorithm into a driving/braking torque command of each wheel hub motor, and entering the next sampling moment And repeating the steps S1 to S3 to form closed loop feedback control.
  9. 9. A distributed driving vehicle stability control system, which is applicable to any one of the distributed driving vehicle stability control methods of claims 1-8, and is characterized by comprising a data acquisition and control triggering module, a time lag compensation model building module and an optimal control and execution module; The data acquisition and control triggering module is used for acquiring real-time state data of the vehicle, calculating a transient load transfer index, calculating time lag compensation time of an actuator based on a secondary transient instability early warning algorithm, judging whether to trigger control intervention, sending a starting signal to the time lag compensation model building module and the optimal control and execution module if the time lag compensation intervention is triggered, and continuously acquiring the data and monitoring the state of the vehicle if the time lag compensation model building module and the optimal control and execution module are not triggered; the time lag compensation model construction module is used for receiving the starting signal and the real-time state data of the vehicle sent by the data acquisition and control triggering module, constructing a nominal dynamics model containing the time lag of the actuator, learning residual errors between actual vehicle dynamics and the nominal model by using a sparse Gaussian process, constructing a data-driven time lag compensation model, and sending the time lag compensation model to the optimal control and execution module; the optimal control and execution module receives a starting signal of the data acquisition and control triggering module and a time lag compensation model sent by the time lag compensation model construction module, adopts an anti-time lag optimal control strategy based on the time lag compensation model, constructs and solves a constraint optimization problem in a limited time domain through a random model predictive control framework, and sends a control instruction to a vehicle executor based on an optimal control input obtained by solving, so as to realize transient stability control of the vehicle.
  10. 10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores a program that, when run on a terminal device, performs the distributed drive vehicle stability control method according to any one of claims 1 to 8.

Description

Distributed driving vehicle stability control method, system and medium Technical Field The invention relates to the technical field of intelligent control of vehicles, in particular to a distributed driving vehicle stability control method, system and medium. Background With the deep collaborative integration of heavy vehicle electrification and intelligent networking technologies, transient stability control of a distributed drive semitrailer has become one of the key challenges facing system design. The wide application of the wire control technology, CAN bus communication and distributed actuators introduces unavoidable actuator time lags, which brings serious tests to the vehicle dynamics control system. The time lag causes the state feedback of the controller to be seriously delayed from the real-time dynamic state of the vehicle, so that the traditional controller is difficult to maintain the vehicle posture under the extreme working conditions of high-speed steering or emergency lane change and the like, and even serious traffic accidents such as folding, side turning and the like are caused. Therefore, research on transient stability control of a distributed driving semi-trailer considering the influence of time lag of an actuator to improve driving active safety has become a problem to be solved urgently. Although the prior art attempts at model-based feedforward compensation, effective coping strategies are lacking for the strong nonlinear characteristics of the distributed drive semitrailer and the random time-lapse variability and the model mismatch problems caused by the same, and it is difficult to ensure the robustness and reliability of vehicle control under complex and variable time-lapse working conditions. Therefore, how to provide a distributed driving vehicle stability control method, system and medium is a problem that needs to be solved by those skilled in the art. Disclosure of Invention In view of the above, the present invention provides a method, a system and a medium for controlling stability of a distributed driving vehicle, which aims to solve the above technical problems. In order to achieve the above purpose, the present invention adopts the following technical scheme: A distributed drive vehicle stability control method comprising the steps of: S1, collecting real-time state data of a vehicle, calculating a transient load transfer index, and calculating time lag compensation time of an actuator based on a secondary transient instability early warning algorithm so as to judge whether control intervention is triggered or not; S2, constructing a nominal dynamics model containing actuator time lag, and utilizing a sparse Gaussian process to learn residual errors between actual vehicle dynamics and the nominal model to construct a data-driven time lag compensation model; S3, based on a time lag compensation model, adopting an optimal control strategy for resisting time lag, adopting a random model prediction control framework, constructing and solving a constraint optimization problem in a limited time domain, and realizing transient stability control of the vehicle based on optimal control input. Further, the collecting real-time state data of the vehicle, and calculating the transient load transfer index specifically includes: Vehicle state data are collected in real time, and the transient load transfer index TLTI value at the current moment is calculated according to the following formula: (1), wherein T represents the track width of the vehicle, , , , , Of the above-mentioned parameters,For the corresponding unsprung mass of each wheel,For vertical acceleration of the unsprung mass of each wheel,The roll moment of inertia of the tractor and the semitrailer respectively,The roll acceleration of the tractor and the semitrailer respectively,In the form of a sprung mass,G is the total mass of the whole vehicle, g is the gravity acceleration,As the vertical acceleration of the sprung mass,The heights of the centers of gravity of the tractor and the semitrailer respectively,Lateral acceleration of the tractor and the semitrailer, respectively. Further, the calculating the time lag compensation time of the actuator based on the secondary transient instability early warning algorithm specifically comprises the following steps: Defining a short-term prediction equation: (2), Calculating TLTI the first derivative and the second derivative by using a difference method, and calculating the residual time required for reaching a preset instability threshold from the current moment, namely the time lag compensation time ADCT of the actuator: (3)。 Wherein TLTI is the transient load transfer index. Further, the step S1 further includes: when the calculated actuator time lag compensation time ADCT is less than the critical activation threshold When the system is in imminent instability, the trigger zone bit is immediately set=1, Activating the subsequent optimal control strategy, only when t