An Improved Particle Filtering Algorithm Based on Consensus Fusion Sampling

Particle filtering is briefly introduced first. Because the depletion of particle diversity resulted from re-sampling causes the decline of filtering precision, an improved particle filtering algorithm based on consensus fusion sampling is proposed. After the re-sampling process, the new algorithm extracts candidate particles based on Markov Chain Monte Carlo (MCMC) principle and combines the re-sampling particles to construct a candidate particle set. Then according to the principle of analytic hierarchy process (AHP), consensus matrix is established, and the complementary and redundancy information of the candidate particles is fully used. Finally, the optimal selection of particles is realized by calculating consensus matrix. Simulation results show the method can effectively reduce the phenomenon of particle impoverishment and improve the state estimation precision.