Lock-in Thermography and Machine Learning-Based Defect Detection in Carbon Fiber Reinforced Thermoplastics

This study evaluates Lock-in Thermography (LiT) for non-destructive defect detection in Carbon Fiber Reinforced Thermoplastics (CFRTP) with Polyphenylene Sulfide (PPS), Polyamide 6 (PA6), and Polyetheretherketone (PEEK) matrices. Artificial delaminations were introduced using Kapton tape, and phase contrast variations were analyzed at different excitation frequencies. A decision tree model was applied to classify defects from LiT data, achieving high accuracy. Ultrasonic C-scan validation confirmed the effectiveness of LiT, though limitations were observed in PEEK samples due to high processing temperatures. The results demonstrate LiT’s potential for CFRTP inspection, with machine learning enhancing automated defect classification.