Adaptive fault estimation for cyber-physical systems with intermittent DoS attacks

Abstract

Abstract This paper is concerned with the problem of fault estimation (FE) for cyber-physical systems (CPSs) modeled by interconnected systems. Under denial-of-service (DoS) attacks, the measurement transmission over the communication network is interrupted, which leads to the infeasibility of the existing FE methods. To overcome this difficulty, a novel switching-type FE scheme is proposed. More specifically, a sequence of adaptive observers and an algorithm are constructed to provide the estimates with improved estimation accuracy, and the observer gain matrices are switched to be zero to discard the outdated measurements under DoS attacks. Therefore, the estimation performance is effectively enhanced in the normal case, and a quantitative description of the estimation error divergence rate is given in the attack status. Compared with the existing results, the interconnections among subsystems are fully considered to improve the estimation accuracy, and in the disturbance-free case, it is proven that the mean sequence of the estimate errors converges to zero. Finally, some simulation results are provide to verify the theoretical findings.