Distributed Dirve Electric Vehicle State-parameter Joint Estimation

Vehicle state and vehicle inertia parameters are the key reference signals for vehicle active safety control systems,such as vehicle horizontal and longitudinal speed,vehicle mass,vehicle center of mass position,and yaw moment of inertia.Real-time and accurate acquisition of state or parameters is to achieve accuracy A prerequisite for active safety control.However,due to the high price of some sensors used for vehicle state measurement,it is difficult to be widely used.Therefore,in general,the methods of obtaining these states or parameters adopt online estimation methods.The existing estimation methods generally use Kalman filter estimation.The main work of this paper is as follows:(1)The introduction of tire force was avoided in the process of building the estimation model.According to the characteristics of distributed driving electric vehicles,the tire force can be calculated by the state of the vehicle itself.This method is verified on the simulation platform.(2)A multi-scale framework based on dual unscented Kalman filters is proposed,which separates the time scales of two unscented Kalman filters,so that the update frequency of the parameter estimator is lower than that of the state estimator.In addition,a state projection scheme is designed to realize the information transfer between the two estimators under different time scales.(3)In the state estimator,under the premise of introducing the sliding window mechanism,the maximum posterior probability estimation and the unscented smoothing method are used to estimate the noise covariance matrix,and it is considered that the fixed window cannot be considered in real-time estimation.For steady-state error and dynamic response,a window length adjustment strategy is used.(4)Considering that the inertia parameters of the car can be regarded as time-invariant when the car is running,this paper designs the start and stop conditions of the parameter estimator,and closes the estimator when the parameter estimator converges to near the true value.Reduce unnecessary calculation load,and use longitudinal velocity as the trigger threshold of the parameter estimator.Finally,this paper has carried out simulation platform verification and experimental verification on the above-mentioned main work.The simulation results and experimental results show that it has good robustness and high accuracy.