Particle filter based on improved genetic algorithm resampling

Abstract

For solving the problem of sample impoverishment in particle filter resampling, this paper proposes a particle filter based on improved genetic algorithm resampling combined with characteristics of selection operator, crossover operator and mutation operator in the genetic algorithm. In the improved genetic algorithm, we choose the importance weight of particles as the fitness value, select particles by utilizing simple resampling and elitist selection, and conduct crossover and mutation operation according to the changeable crossover probability and changeable mutation probability based on the degree of particle degeneracy. Simulation results demonstrate that the particle filter algorithm based on the improved genetic algorithm resampling could guarantee the validity of the particles and increase the diversity of the particles. This algorithm could improve the performance of the particle filter.