Interacting Multiple Model Algorithm Using Skew-t Measurement Noise and Variational Bayesian Approach

摘要

Based on linear discrete-time state-space models with Skew-t measurement noise and Gaussian process noise, this paper proposes an interacting multi-model (IMM) filtering algorithm. This algorithm uses a variational Bayesian approach to approximate the posterior distribution of models. In simulations, the proposed IMM algorithm is compared with several single model filtering algorithms using variational Bayesian approach, which shows the superiority and effectiveness of this IMM algorithm.