Weight degeneracy is a grievous defect for particle filter which result in wasting large computation cost. Although re-sampling can solve degeneracy issue in some extent, but it lead to the particle impoverish phenomenon which weaken Bayesian inference precision for state estimation. A new particle filtering algorithm was proposed, which introduced Swarm Intelligence niching technique into particle filtering in order to enhance particles diversity and search ability and convert above issues into a multi-modal optimization so that measure information is fully. As a result, particles can move towards the high likelihood areas which increase precision for system state estimation. Simulation results show that the proposed particle filtering is effective and stable.