Adaptive IMM Filters for Maneuvering Target With Additive and Multiplicative Noises

Compared with the existing interactive multiple-mode (IMM) filters, this paper focuses on IMM filter with unknown additive and multiplicative noise covariances. The adaptive IMM (AIMM) filters to deal with the unknown additive noise covariance matrix and multiplicative noise variance are derived in the paper. First, the noise covariance is characterized by an inverse Wishart (IW) prior, and the target state is modeled as a Gaussian IW (GIW) distribution. Then the GIW implementations of the proposed AIMM filters based on variational Bayesian inference are given, and the mixing of GIW components is calculated according to the weighted Kullback–Leibler average. Finally, simulation and experimental results demonstrate that the proposed AIMM filters can estimate maneuvering target state effectively.