| Distributed drive electric vehicle is an ideal carrier for automatic driving in the future.However,the new chassis structure of distributed drive electric vehicle also brings new challenges to the control of vehicle handling and stability.For the integrated control of vehicle handling and stability,the first goal is to obtain the key state information of the vehicle.Because the four wheels of distributed drive electric vehicle can be driven and braked,the method of estimating the longitudinal speed using the non-driving wheel speed is no longer applicable.The yaw rate and the slip angle of vehicles obtained by inertial measurement unit(IMU)is not accurate due to the deviation of sensors.In addition to the problems in state estimation,there are some shortcomings in the design of integrated controller.For example,the upper level nonlinear controller in the hierarchical control structure has complex structure,and the uncertainty modeling is not considered enough.These factors restrict the practical application of distributed drive electric vehicles.In order to solve the above problems,a vehicle state observer and an integrated controller are designed in this paper.The main contents are as follows:(1)According to the requirement of state estimation and integrated control,kinematic relations of coordinates of positioning system,inertial coordinates and vehicle coordinates are established.Vehicle dynamics models are also established.The magic tire formula parameters are fitted by Carsim’s actual tire data.(2)The state of vehicle is estimated based on intelligent tire sensor,Inertial Navigation System(INS)and Global Navigation Satellite System(GNSS).The longitudinal speed is estimated by information fusion of the wheel speed,the GNSS velocity and the longitudinal acceleration.A weighted fusion Kalman filter is designed to reduce the dimension of the observation matrix.The adaptive adjustment method of observation noise variance based on the reliability index function is designed according to the prior knowledge.In order to estimate the yaw rate and the side angle,the GNSS velocity and azimuth are used to correct the deviation of the yaw rate and the lateral acceleration in real time.The adaptive adjustment of the variance of observation noise is realized by the improved Sage-Husa adaptive Kalman filtering algorithm.(3)The integrated controller of vehicle adopts a hierarchical control structure.To obtain the generalized longitudinal force,lateral force and yaw moment,the upper controller is based on the unconstrained model predictive control algorithm.Meanwhile,a nonlinear disturbance compensator is designed.The lower level controller establishes the control allocation objective with the minimum generalized force error and the maximum tire stability margin.Actuator saturation constraints are considered.The tire friction ellipse constraint is fitted by the twelve edges.The generalized force and torque are assigned by solving the optimization problem.Aiming at the failure of single actuator,a method based on restricted memory least squares is designed for fault detection and identification.Combined with fault information,the control allocation law is reconstructed to achieve vehicle fault tolerant control.(4)The simulation platform of Carsim and Matlab/Simulink is used to verify the effect of the state observer and controller.The simulation results show that the proposed state estimation algorithm can achieve accurate estimation of the longitudinal speed at constant speed,driving and emergency braking situation.The GNSS signal can correct the yaw rate deviation and lateral acceleration deviation,which improves the estimation precision of sideslip angle.The integrated controller can deal with uncertainty modeling and reduce the tire utilization and improve vehicle handling in normal conditions.The controller can also ensure the stability of the vehicle under the limited condition.In the case of actuator failure,the direction stability of the vehicle will be maintained and the steering ability of the vehicle is guaranteed.(5)The actual estimation effect of the state observer is verified by the real vehicle acquisition data.The results show that the proposed method can correct the deviation of the actual INS sensor,and can accurately estimate the state of the vehicle under the condition of straight driving and cornering. |
