A distributed fusion filter based on simultaneous estimation of state and inputs

In this paper, for a discrete random linear system with unknown input, a three-step recursive filter for linear feedthrough with unbiased estimation of unknown inputs and states is considered without any prior information of unknown inputs. When the system has multiple sensors, the mutual variance matrix of the filtering error between any two sensor sub-systems could be deduced, and then the weight of each sensor state component is given according to the scalar weighted fusion of the multi-sensor linear minimum variance component Distributed filtering fusion. The simulation results verify the effectiveness of the proposed method.