Data-driven Distributed Consensus Filter for a Discrete-time Nonlinear Sensor Network

In this work, a novel data-driven distributed consensus filter (DD-DCF) is proposed based on the dynamic linearization technique (DLT) for a discrete-time nonlinear sensor network. Compared with conventional model-based consensus filters, the proposed method is data-driven merely depending on the input and output (I/O) data from measurements. Both the data-driven system identification (DD-SI) algorithm and the distributed consensus filter state estimation (DCF-SE) algorithm are investigated for a nonlinear sensor network. The theoretical analysis shows the main result of the DD-DCF algorithm in the sensor network. The simulation results verify the effectiveness of the designed approach.