Fault diagnosis in complex systems using artificial neural networks

Very complex technical and other physical processes require sophisticated methods of fault diagnosis and online condition monitoring. Various conventional techniques have already been well investigated and presented in the literature. However, in the last few years, a lot of attention has been given to adaptive methods based on artificial neural networks, which can significantly improve the symptom interpretation and system performance in a case of malfunctioning. Such methods are especially considered in cases where no explicit algorithms or models for the problem under investigation exist. In such problems, automatic interpretation of faulty symptoms with the use of artificial neural network classifiers is recommended. Two different models of artificial neural networks, the extended backpropagation and the radial basis function, are discussed and applied with appropriate simulations for a real world applications in a chemical manufacturing plant