A Study on Method of Intelligent Fault Diagnosis in Large and Complex Device

Abstract

Recently neural network is widely used in fault diagnosis. As the neural network method has the disadvantages of the slow convergence rate and the uncertain node number in neural network hidden layer, the results of fault diagnosis in complex devices are not satisfactory. This paper combines fuzzy logic and neural network, and presents a multilayer feed-forward fuzzy neural network with a serial structure for fault diagnosis. We take the fault diagnosis in ship diesel engine as an example to simulate. The results demonstrate that the method could improve the speed of diagnosis greatly and could diagnose and predict the device faults timely and rapidly. It can significantly improve the fault diagnosis to apply the method in large and complex devices.