State estimation using a fuzzy adaptive particle filter for nonlinear stochastic systems
Takoua Grami, Ali Sghaïer Tlili · 2017
Estimation of nonlinear stochastic systems is one of the important issues discussed by researchers. Indeed, the particle filter has emerged as a consistent technique to study nonlinear systems, and has been considered as a defiance issue relying on a proficient distribution allowing consequently an efficient number of particles. In this paper, a recursive estimation algorithm using particle filtering based upon the fuzzy logic concept is rigorously developed and demonstrated for nonlinear stochastic systems. The fuzzy logic approach is introduced to calculate the number of particles that will be changed in each iteration by using the difference between the actual and the estimated states for reliable estimation. Further benefits of the proposed approach are especially the performance improvement and time optimization with respect to the standard particle filter (SPF). The efficacy of the proposed fuzzy adaptive particle filtering (FAPF) approach is highlighted via numerical simulation applied on a nonlinear stochastic system represented by the widespread Holmes map process. In this context, a comparative study of performances with the standard particle filter is carried out.