Contaminative Data-Driven Koopman Resilient Distributed Filtering for Unknown Stochastic Nonlinear Systems.

Abstract

This article proposes a Koopman-enhanced distributed filtering for unknown stochastic nonlinear systems using contaminated datasets. To overcome the limitation that conventional Koopman operators fail to handle unknown stochastic dynamics, a delay-coordinate embedding strategy is introduced to reconstruct the lifting observations from noisy measurements. Moreover, to suppress the adverse effects of process and measurement noise on the invariant subspace, a robust subspace dynamic mode decomposition (SDMD) method is developed for reliable Koopman operator identification. Within this framework, a distributed filtering scheme is designed that exploits both direct and indirect measurements, where an adaptive event-triggered mechanism is further derived to balance network transmission burden and estimation accuracy. Finally, simulation results demonstrate the effectiveness and robustness of the proposed filtering approach.