Failure-tolerant performance stabilization of the generic large-scale system by decentralized linear output feedback

Most of the existing results on the decentralized stabilization of large-scale systems consider systems in the input-output-decentralized form, i.e., systems whose input and output matrices are block diagonal. The controller synthesis for such systems is less involved than that for input-output-centralized ones. In this paper, a transformation is proposed for the input-output decentralization of the generic large-scale system. Using the transformed system, a sufficient condition for failure-tolerant performance stabilization of the original large-scale system in a desirable performance region under decentralized linear output feedback is established. The problem is then reformulated as a constrained nonlinear optimization problem. The proposed methodology results in the optimal reconciliation of failure-tolerant performance stabilization of the overall system, and reliability and low actuator gains of the isolated subsystems. The effectiveness of the proposed approach is demonstrated by an example.