Robust steady‐state matrix‐weighted fusion filtering for multisensor multichannel autoregressive signal with multiple uncertainties

Abstract

SummaryThis article investigates the robust steady‐state matrix‐weighted fusion (MWF) filtering problem for multisensor multichannel autoregressive (AR) signal with multiple uncertainties, which can be transformed into the state estimation problems in the state space representation. The uncertainties include the random parameter matrices, uncertain noise variances (UNVs), one‐step random measurement delay, missing measurements, and packet dropouts. Three Bernoulli distributed stochastic variables with known conditional probabilities are introduced to describe the networked stochastic uncertainties. The AR signal model under consideration is transformed into a state space model only with UNVs via using a combined method, which is composed of the state space method, the augmented method, and the fictitious noise method. Based on the worst‐case system with conservative upper bounds of UNVs, the robust local steady‐state one‐step and multi‐step signal predictors are proposed by using the minimax robust estimation method. Subsequently, the distributed fusion signal predictors weighted by matrices are addressed. The robustness of the proposed signal predictors is proved by the introduced augmented noises and permutation matrices. Simulation experiment demonstrates the correctness and effectiveness of the proposed methods.