Online Fault Diagnosis of Dynamical Systems in a Collaborative Sensor Network

In this contribution, we propose an efficient collaborative strategy for online change detection, in a distributed sensor network. The collaborative strategy ensures the efficiency and the robustness of the data processing, while limiting the required communication bandwith. The observed system is assumed to have a finite set of states, including the abrupt change behavior. For each discrete state, the system is assumed to evolve according to a linear state-space model. An efficient Rao-Blackwellized collaborative particle filter (RB-CPF) is proposed to estimate the probability of occurrence of the system state change, within a Bayesian framework. The Rao-Blackwellization procedure combines a sequential Monte Carlo filter with a bank of distributed Kalman filters. The sensor network has a circular architecture where each smart sensor is able to exchange sufficient statistics and soft decisions with its two nearest neighboring sensors. The local fusion is based on the selection with replacement (resampling) algorithm.