A Novel Particle Filter Aided by Interval Analysis Approach

This paper addresses the issue of state estimation in nonlinear systems in the presence of non-Gaussian and bounded noises, under which the interval analysis based estimation is introduced as an auxiliary approach of generic particle filter (PF). This yields the so-called Set-Membership aided particle filter (SMPF). Unlike the mature alternatives of generic particle filter, the proposal distribution of SMPF approximates the posterior probability density function (PDF), not only on the numerical value but also on the definition-domain, and the performance analysis on the proposed alternative is proven through detailed formulations. In addition, contrasting simulations under SMPF and other mature alternatives also validate the effectiveness of SMPF.