Design considerations for a motor fault detection artificial neural network

The authors discuss the design considerations for a motor fault detection artificial neural network in terms of determining the input/output training data, the size of the training data set, network accuracy, robustness, implementation feasibility, and the number of input and hidden nodes to be used. A fuzzy logic approach to automating the network configuration process while simultaneously considering the accuracy, training time, sensitivity, and the number of neurons used in the implementation is also presented. Successful results have been obtained using artificial neural networks for motor fault detection and fuzzy logic in the network configuration design. A feedforward neural network for performing fault detection in a split-phase squirrel-cage induction motor is used for illustration purposes