Self-Tuning Information Fusion Reduced-Order Kalman Predictors for Stochastic Singular Systems

Using correlation functions, a distributed identification approach for noise statistic information is given for stochastic singular systems measured by multiple sensors with unknown noise statistics information. Compared with the centralized identification method, the computation burden can be reduced. Further, a self-tuning information fusion reduced-order Kalman predictor with a two-stage fusion structure is presented based on the fusion algorithm weighted by scalars in the linear minimum variance sense. The first stage fusion is to determine the correlated variances of measurement noises between any two sensors. The second stage fusion is to obtain the distributed self-turning information fusion reduced-order predictors by scalar weighting fusion based on local predictors from each sensor subsystem. Simulation example shows the effectiveness of the proposed algorithm