On-Line Fault Detection by using Filters Bank and Artificial Neural Networks

The aim of this paper is to propose a method for the detection of faults in industrial systems, such as electrical machines and drives, through on-line monitoring system. Early fault detection, which reduces the possibility of catastrophic damage, is possible by comparing the measured signals with a database that contains characteristic signals for machines operating with and without faulty conditions. This approach is based on a Filters Bank that extracts frequency and energy characteristic features, and an Artificial Neural Networks (ANN) that classifies these features. A link with the wavelet transform for features extraction is also presented. The faults that are concerned correspond to a change in frequency components of the signal.