Multi-Sensor Distributed Fusion Filter for Stochastic Singular Systems with Unknown Input

Based on decomposition in canonical form, a singular system is transferred into two equivalent reduced-order subsystems for a single-sensor stochastic singular system with unknown input. Without any prior information of the unknown input, a reduced-order state filter in linear unbiased minimum variance sense is presented, which is independent of the unknown input. Further, based on the scalar-weighted fusion algorithm in the linear minimum variance sense, a multi-sensor distributed information fusion state filter is given for multi-sensor stochastic singular systems with unknown input. The filtering error cross-covariance matrix is derived between any two local estimators. The simulation research verifies its effectiveness.