Robust Multilayer Parallel Covariance Intersection Fusion Predictor for MultiChannel AR signal with Colored Noises and Uncertain Noise Variances and Unknown Cross-Covariances

This paper is concerned with robust multilayer parallel covariance intersection fusion prediction problem for a class of multisensor multichannel AR signal systems with colored noises, uncertain noise variances and unknown cross-covariances. By the state space method, the original system is converted into a multi-model system with only uncertain noise variances and unknown cross-covariances. According to the mini-max robust estimation principle, based on the worst-case system with conservative upper bounds of the noise variances, the local robust Kalman signal predictors are presented. And the MPCI fusion predictor is presented according to the MPCI fusion algorithm presented in this paper. The robustness of the proposed local and MPCI fused predictors are proved such that their actual prediction error variances are guaranteed to have the corresponding minimal upper bounds for all admissible uncertainties. The robust accuracies among the local and MPCI predictors are proved and they are compared by the traces of the variance matrices and the covariances ellipsoids, respectively. Finally, a simulation example shows the correctness and effectiveness of the proposed results.