Subspace identification for distributed fault diagnosis of large-scale interconnected systems

Abstract

For data-driven fault diagnosis of large-scale interconnected systems, if it is treated as a centralized system, it will lead to a large time complexity of the algorithm and large storage resources. Aiming at this problem, this paper proposes a distributed fault diagnosis method. Firstly, the large-scale interconnected system is equivalent to several independent subsystems by using algebraic equivalence transformation. At this time, the unmeasurable input will not be contained in all local networks, that is, the unknown interaction item is decoupled in the equivalent system. Then, a subspace identification-based approach is applied to these independent subsystems to construct the residual generator directly for each subsystem without the need to identify system parameters for distributed fault detection, isolation and identification. A numerical simulation example is given to verify the effectiveness of the proposed method.