On the application and design of artificial neural networks for motor fault detection. II

The emerging technology of artificial neural networks has been successfully used in a variety of different areas such as fault detection, control, signal processing, and many others. This paper presents the general design considerations of feedforward artificial neural networks to perform motor fault detection. The paper first discusses a few noninvasive fault detection techniques, including the parameter estimation approach, human expert approach, etc., and then the artificial neural network approach. A brief overview of feedforward nets and the backpropagation training algorithm, along with its pseudo codes, will follow. Later sections explain some of the neural network design considerations such as network performance, network implementation, size of training data set, assignment of training parameter values, and stopping criteria. Finally, a fuzzy logic approach to configure the network structure is presented.