Distributed state estimation and fault diagnosis for networked systems under jointly connected switching networks

In this paper, the distributed state estimation of the target system and the distributed fault diagnosis of the sensor network are investigated for networked systems under jointly connected switching networks. In this scenario, to eliminate the effect of external disturbance, we first construct a distributed observer system consisting of N distributed unknown input observers (UIOs), each of which is located on the sensor and can only access the partial input and output information of the target system and the transmitted state estimation from its neighbors. Owing to the limitation of communication conditions, the network topology of the distributed observer system is often time-varying and disconnected. To solve this problem, a novel observer parameter design method is proposed, which overcomes the limitation that the existing techniques can only rely on fast switching to estimate the state of the target system. On this basis, a distributed sensor fault diagnosis scheme based on the observer system is designed and its performance is analyzed. In addition, a faulty sensor replacement strategy based on graph theory and system detectability is proposed to ensure that the faulty sensors will not affect the state estimation performance of the distributed observer system. Finally, an illustrative example is introduced to verify the effectiveness of the proposed scheme.