Machine learning is a branch of Artificial Intelligence that is used effectively and reliably to determine the solution to a problem. In deciding on the intensity of artefacts for different industrial applications, identification of subsurface non-uniformity is essential. In order to detect anomalies in a wide variety of materials, non-stationary thermal wave imaging is emerging as a reliable qualitative evaluation technique. This paper proposes a classification modality based on supervised machine learning to detect subsurface defects using modulated quadratic frequency. With 10 Teflon patches having different depths and sizes, thermal wave imaging and experimentation was carried out over glass fibre reinforced polymer material (GFRP) and carbon fibre reinforced polymer (CFRP) with 25 bottom holes with varying sizes and depths. Three well-known supervised machine learning techniques are used to detect defects in this paper Decision tree (DT), Support vector machine (SVM) and k-nearest neighbour (KNN) classifiers. The detection capability and reliability of the detection of defects were evaluated using the signal to noise ratio and the detection probability, respectively. Decision tree classifier offers better detection capability, sizing and reliability estimation compared to remaining processing methods among various supervised learning methods.

Author (s) Details

Mrs. A. Vijaya Lakshmi
Infrared Imaging Center, K L E F University, Vaddeswaram, 522502, India.

Dr. V. S. Ghali
Infrared Imaging Center, K L E F University, Vaddeswaram, 522502, India.

View Book :-

Leave a Reply

Your email address will not be published. Required fields are marked *