The ant system-genetic algorithm particle filter
Juan Zhao, Dongfeng Li · 2010
Particle filter is a statistic filtering method based on sequential simulation. It has an outstanding contribution to the nonlinear non-Gaussian dynamic system. But how to choose particle probability distributing function and deal with particle degeneration is the key to the algorithm. A new evolutional algorithm called ant system is used during the iterative recurrence of sequential important sampling. Furthermore, the particle diversity was great increased by the using of genetic across, aberrance and selection. Simulation results show that this evolutional is better than traditional particle filter in the average absolute error and variance within a short time.