State and parameter estimation based on a modified particle filter for an in-wheel-motor-drive electric vehicle

This paper presents a modified particle filter (MPF) to estimate vehicle states and parameter with high precision and robustness under complex noises and sensor fault conditions. To deal with the particle impoverishment issue, the vector particle swarm of the multivariable system is separated into univariate particle swarms, which are diversified with the selection, crossover and mutation operations of the genetic algorithm (GA) while maintaining the mean value and enlarging the standard deviation. The effectiveness of the proposed estimation scheme is verified under the scenarios of the stochastic and needling noises and acceleration sensor faults through the Carmaker-Simulink joint simulations based on typical maneuvers, outperforming the commonly-used vehicle state estimators including the unscented Kalman filter (UKF) and the unscented particle filter (UPF).