Decentralized fault accommodation of a class of interconnected nonlinear systems using an adaptive approximation approach

This paper presents a decentralized adaptive approximation design for the fault tolerant control of interconnected subsystems. We consider faults that occur in the subsystems local dynamics as well as in the interconnection effects. Linearly parameterized neural networks are used to adaptively approximate the unknown interconnection effects and the unknown changes in model dynamics due to failures. An adaptive bounding method is introduced for addressing stability and robustness issues in the presence of residual approximation errors. A simulation example is presented to illustrate the proposed control design methodology.