Incipient Fault Detection In DC Machines Using A Neural Network

Abstract
Rotating machines are used extensively to convert electrical energy to mechanical energy. In this paper, an incipient fault detection scheme using a neural network is presented. An incipient fault within a motor usually affects its performance. Factors such as conductive contamination, loose hardware, design defects, worn bearings, abnormal operation, aging and disturbance can cause damage to the motor. The symptoms of the faults are usually reflected by changes in parameter values and the steady-state operating condition of the motor. The neural network is basically a software (or hardware) implementation of parallel computing which identifies different input patterns and generates appropriate outputs. Data is collected on the parameters and state values of the motor under different fault conditions. Based on these data, a neural network is then implemented to perform incipient fault detection.

This publication has 4 references indexed in Scilit: