Abstract:

An improved particle filter based on the ensemble Kalman filter (EnKF) is described. The EnKF is used to propagate the particle’s proposal distribution of particle filter in every time step. Because the EnKF can attain a maximum posterior estimation of the nonlinear system, and the proposal distribution function integrates the latest observation into system state transition density, so the proposal distribution can approximate the true posterior distribution more accurately. This new algorithm uses sample ensembles to approximate the nonliner proposal distribution, and the number of ensembles required in the EnKF is flexible. 〖JP2〗In this way, the accuracy and efficiency of the algorithm is improved. The simulation results reflect that the novel particle filter is superior to the standard particle filter, the extended Kalman particle filter and the unscented particle filter.