| As a new type of vehicle,the distributed drive electric vehicle is regarded as the main development direction of future vehicles,because each wheel is directly controlled by an independent drive motor,which lead to rapid response,accurate control and other advantages.How to achieve a higher level of driving safety and intelligent driving based on electric vehicles has become an urgent problem.Intelligent driving assistance systems(ADAS)is an important guarantee of vehicle driving safety,and accurate and real-time vehicle state parameter information and road information is the premise of control system to achieve accurate decision-making and control.Although the distributed drive electric vehicle can obtain accurate wheel torque and speed information through the drive motor without adding sensors,many necessary parameters can not be obtained directly or the cost is too high.Therefore,it is important to estimate the required parameters by using reasonable algorithm based on the existing sensor measurement information.Firstly,according to the technical parameters of a real vehicle,the co-simulation model of distributed electric drive vehicle based on MATLAB/Simulink and Truck Sim is established,and the effectiveness of the model is verified by the simulation of typical working conditions.Considering the influence of the time-varying characteristics of tire cornering stiffness on the estimation of vehicle state parameters,an on-line updating estimation algorithm of tire cornering stiffness using extended Kalman filter and forgetting-factor recursive least square method(EKF+FFRLS)is proposed to estimate the longitudinal speed,yaw rate and side-slip angle of electric vehicles,and the results are compared with the standard extended Kalman filter(EKF)estimation results.Forgetting-factor recursive least square method(FFRLS)is used to estimate the change rate of road slope,which is used as the input information of EKF estimator to jointly estimate vehicle mass and road slope.Then,the above estimation algorithms are integrated and designed.At the same time,fuzzy control rules are established to adjust the Kalman gain matrix of state estimation in real time,and a robust overall system with joint estimation of vehicle state parameters and road slope and mutual feedback of parameters is constructed.Finally,the effectiveness of the algorithm is verified based on the co-simulation vehicle model.The research shows that:(1)The proposed EKF + FFRLS algorithm estimates the tire cornering stiffness as a time-varying parameter,which makes up for the lack of modeling accuracy and improves the estimation accuracy of vehicle state.As far as the side-slip angle of vehicle is concerned under high-speed and full-load condition,the mean absolute error and root mean square error of EKF + FFRLS algorithm are 74.6%and 76.4% lower than that of EKF algorithm,respectively;(2)Taking the estimated value of road slope change rate as the input signal of the slope estimator can effectively improve the phenomenon of slope estimation delay,and the mean absolute error of vehicle mass and road slope estimation is reduced by 49.2% and 82.1% respectively compared with that without considering the slope change rate.(3)The proposed vehicle state estimation algorithm is organically combined with the mass and slope estimation algorithm,and the fuzzy controller is introduced to adjust the Kalman gain matrix of EKF + FFRLS algorithm in real time to form an integrated estimation algorithm.Compared with EKF + FFRLS algorithm,the maximum relative errors of yaw rate and sideslip angle obtained by the integrated estimation algorithm are reduced by 4.5% and35.7% respectively when the vehicle is fully loaded.Therefore,the algorithm can better adapt to the changes of input signal with time-varying noise and vehicle load,realizing the adaptive effect of the algorithm. |
PDF Full Text Request