Automated Eddy Current non-destructive testing through low definition lissajous figures

In the framework of Eddy Current Testing (ECT), this work presents an automated non-destructive testing method based on Eddy Currents which uses few geometric features of low definition Lissajous figures. A features vector representing the shape of the ECT magnetic field response represented in the complex plane is used as signature to recognize specific defects of aerospace structures. In order to evaluate the proposed method, the accuracy, specificity, sensitivity, precision, F-Measure, AUC, and Matthews correlation coefficient are used to rate the following classifiers: J48, multi-layer neural network and Naive Bayes. The used data set is carried out by lab experiments performed on an aircraft test-piece with several well-known defects. The results show the usefulness of the proposed approach to be used as an aided tool for ECT data analysis. It allows both an easier and shorten data interpretation by the qualified inspectors, and an increase in the diagnosis quality.