A Wavelet-Based Method for Defect Detection in Aircraft Components by Using Ultrasonic Technology

Manufacturing processes in the aerospace context, although strictly supervised and inspected, can sometimes create internal defects in the final components. This paper presents a wavelet-based method for the non-destructive detection of these unwanted defects. By opportunely inspecting the components using laser ultrasonic technology, it is possible to detect defective parts. The wavelet analysis allows us to extract significant and distinctive features (signatures) from the ultrasonic data. As the computed signatures of defective areas differ from the non-defective ones, it is possible to discern these areas effectively. Different machine learning methods for comparing signatures are proposed. High detection accuracies (>97%) have been achieved by investigating one specimen with six different defects. The obtained results allow us to affirm that the proposed approach looks promising and suitable for this purpose, and thus needs further investigations.