Distributed subspace identification for data-driven fault detection and fault-tolerant control of complex interconnected, industrial systems

In response to the growing demands for enhanced safety and reliability in modern process industry, this paper proposes a distributed-subspace-aided fault detection (FD) and fault-tolerant control (FTC) strategy for complex interconnected, industrial systems. Thus, the distributed input/output (I/O) data models are first derived for the subsystems, within the permutation and decomposition of the centralized I/O data model for the full interconnected system. Building upon this foundation, distributed subspace identification (DSI) is realized, using an oblique projection algorithm and information exchange between the subsystems. This is followed by a distributed FD and FTC strategy for the full interconnected system, where the residual generators and the iterative subspace predictive controllers (ISPC) for the subsystems are designed. Finally, a numerical simulation and a comparative benchmark study are conducted to validate the effectiveness of the proposed distributed SIM-aided FD and FTC strategy. In the numerical simulation, the total integral of absolute error (IAE) of the proposed FTC strategy is 81.5% lower than that without the proposed FTC strategy. In the benchmark study, the total mean squared error (MSE) of the proposed distributed FTC strategy is 47.5% lower than that of the decentralized FTC strategy, which shows the effectiveness of the proposed distributed FTC strategy.