Robustness of an induction motor incipient fault detector neural network subject to small input perturbations

The authors present an incipient fault detector artificial neural network (IFDANN) for single-phase squirrel-cage induction motors and discuss a method for improving the robustness of such a network to small input perturbations for real-time applications. In addition, the concept of input-output sensitivity analysis is used to test the performance of the fault detector neural network with respect to input noise. Simulation results are presented to show the significant improvement in robustness of the modified IFDANN for operation with noisy input measurements. The network modification and the input-output sensitivity analysis presented can be extended to other neural networks designed for online applications, where noise is an important factor