Abstract: The traditional particle filter algorithm is prone to particle depletion and weight degradation, and a large number of particles are needed to make an accurate estimation of non-linear system. A novel particle filter algorithm based on grey wolf optimization algorithm is proposed. The particles are employed to characterize the individuals of grey wolf and simulate the process of wolf hunting, which makes the particles move to the high likelihood region of posterior probability and improve the rationality of the distribution of the particles. Levy flight strategy is introduced in grey wolf algorithm to improve the convergence speed. The adaptive weight adjustment is adopted before partial resampling to increase the diversity of the particles. The results of simulation experiment show that the improved algorithm upgrades the estimation accuracy of particle filter, ensures the diversity of particles and the rationality of particle distribution, and reduces the number of particles required for state estimation.

Key words: particle filter, state estimation, grey wolf algorithm, particle diversity