A comparative study of conventional and artificial neural network classifier for eddy current signal classification

Abstract

A series of eddy current signal trajectories have been obtained on artificial round and rectangular defects in thin stainless steel plates. These signals have been processed to achieve a single waveform characterising the eddy current trajectories from which a large number of features have been derived both in the time and frequency domains. The optimised number of features to characterise a defect has been ascertained with the help of four reported conventional classifiers and an Artificial Neural Network (ANN) classifier. A comparative assessment of the potential of these classifiers has been carried out within the domain of the given signal trainings in this investigation. The results indicate that only five signal features are sufficient to gain an understanding about the nature of defect.