Segmentation of Hidden Delaminations with Pitch–Catch Ultrasonic Testing and Agglomerative Clustering

Abstract

This paper studies the detection of hidden polymer matrix composite delaminations with a pitch–catch ultrasonic testing system and an agglomerative clustering algorithm. Existing ultrasonic testing methods characterize damage through normal-incidence pulse-echo measurements. Yet, these pulse-echo methods are ineffective at detecting delaminations underneath other delaminations. The ultrasonic waves only reflect off the top delamination. As a result, no information about the lower delamination is transmitted to the receiver. To address this problem, we investigate an oblique-angle pitch–catch ultrasonic testing method to transmit ultrasonic waves underneath the delaminations. The ultrasonic waves interact with the lower delaminations and carry that information to the receiver. We describe and discuss our experiments, which use two polytetrafluoro-ethylene (PTFE) inserts to simulate delaminations. We show that applying agglomerative clustering to the experimental data can successfully distinguish three regions: regions with two PTFE inserts, regions with only an upper PTFE insert, and regions with only the lower PTFE insert. Normal-incidence measurements only observe two regions.