Abstract:

The particle filter has become the mainstream method for solving system parameter estimation and the state of filter in nonlinear nongaussian dynamic systems. However the particle degradation problem in particle filter is an inevitable phenomenon and the solution is particle resampling. According to the particle degradation phenomenon of the existing defects, there will be a new mixed particle filter proposed in this paper based on the extended Kalman particle filter. In the new algorithm, the extended Kalman particle filter with support vector machine (SVM) implements the present moment sampling and resampling. This structure makes use of the latest observation information avoiding the lack of particles. It has small errors and better stability. Theoretical analysis and simulation results show that the new method outperform the interacting standard particle filter and the extended Kalman particle filter in the filter precision of doublemodal noise system state.

Key words: particle filter;resampling;SVM;doublemodal noise;