Auxiliary Sequential Importance Resampling Particle Filter (ASIR PF) Based on Particle Swarm Optimization for Nonlinear System StateEstimation

Auxiliary Sequential Importance Resampling Particle Filter is a recursive Bayesian filtering for nonlinear systems with non-Gaussian noise which uses the Monte Carlo method for calculating the posterior probability density functions. In this filter to estimate the system state, the current observations are used to approximate the proposed distribution function and causes particles to be located in areas with a high probability. One problem with this filter and other particle filters that we are facing is the particle degeneracy. Degeneracy phenomenon increases the variance of the weight of the particles after a while thus a divergence in state estimation is created. To minimize this effect, we use Particle Swarm Optimization algorithm which directs the particles toward the greater posterior probability density functions pots.

کلیدواژه ها:

Sequential Monte Carlo ، Probability density function Posterior ، Degeneracy Phenomenon