Adaptive locally-linear-models-based fault detection and diagnosis for unmeasured states and unknown faults

Today the problem of fault detection and diagnosis (FDD) is considered as an important and essential counterpart of control engineering systems. Because of importance and existence of faults that don´t have a known structure in control system, i.e., fault occurred because of tangle of complex factors, In this paper a Lipschitz nonlinear system with unmeasured states and unknown faults is considered and a novel FDD architecture for it is presented. A neuro/fuzzy model consisting of few locally linear models (LLMs) with on-line updated centers and width vectors is used to approximate the model of the fault. A nonlinear observer is used to estimate the states of the system that are inputs to LLMs. The stability analysis of system is carried out via Lyapunov theory, from which the parameter updating rules are derived. At the end of this paper some numerical simulation is given to show the effectiveness of the method.