Method Of Nonlinear Systems Based On Analytical Redundancy Fault Diagnosis

Owing to the increasing demand for high reliability and safety for many modern industrial process, fault detection and isolation(FDI) algorithms and their applications to a wide range of industrial process have been the subjects of intensive investigation over the past thirty years. One of the key issues in the design of fault diagnosis scheme is the effect of modelling uncertainties and disturbances. By using neural networks observer or approximator, the fault detection and diagnosis strategy for a class of nonlinear systems with modelling uncertainties are investigated.The main contents studied in this thesis are as follow:Firstly, the latest development of fault detection and diagnosis is briefly introduced. The fault diagnosis problem for a class of non-linear systems with uncertainty which depends on states, inputs and unknown parameters is discussed. By using the estimations of both states and parameters, the fault can be detected.Secondly, an approach based on neural network estimator for nonlinear system is discussed. The RBF network is used to model non-linear dynamic system by off-line and on-line learning rule. The approach improves the robustness of FDI. In order to further discuss the identification of fault after the detection, a nonlinear online neural network approximator is used to provide an estimate of the fault. On-line approximators and adaptive nonlinear filtering techniques is used to obtain estimates of the fault functions. Robustness, sensitivity and stability conditions of the scheme are rigorously derived.Finally, all the approaches discussed in this thesis are demonstrated through corresponding simulations.