Laser Spot Thermography for Defect Detection in Composite Structure

Abstract

Thermography is a significant part of nondestructive testing and is favorably used in composite materials examination. Specifically, laser thermography, which is one of the active thermography methods, utilizes heat excitation achieved by means of a laser. This technique grants precise and repetitive measurements which contributes to effective data processing and interpretation. In this paper defect detection technique based on active laser thermography measurements is presented. The main part of this work is introducing the processing routine for thermograms’ sequences which ultimately leads to fault detection. Automatic analysis of thermograms utilizing image processing methods allows for straightforward feature extraction for further examination. Having obtained meaningful and concise representations of measurements, machine learning techniques are used for classification purposes. A multi-layer perceptron, a classical neural network, and support vector machine are applied for this task. Finally, a specific area of the examined sample is indicated as faulty or flawless. The method is presented in the case of the composite plate containing prepared defects imitating delamination at various depths.