Abstract: Particle filter based on particle swarm optimization (PSO-PF) algorithm suffers from low precision and high computation complexity, therefore is difficult to be used in practical applications. This paper proposes   novel dynamic particle filter algorithm based on neighborhood adaptive particle swarm optimization (DPSOPF). The method takes the neighborhood information of particles into consideration. Factors of diversity, neighborhood extension, and neighborhood limiting are jointly used to realize self-adaption of neighborhood particle numbers. Thus the influence of particles on the neighborhood is under control, and the local optimization phenomenon is alleviated. Optimal balance is achieved between convergence speed and search ability. By using the univariate nonstationary growth model (UNGM), target tracking model and failure detection model, a simulation test of the algorithm is performed. The results show that, compared to PSO-PF, the proposed algorithm improves precision and computation speed, showing its applicability to practical engineering.

Key words: particle filter, particle swarm optimization, target tracking, fault detection