Autonomous Monitoring of Breathing Debonds in Bonded Composite Structures Using Nonlinear Ultrasonic Signals

This study aims to develop a structural health monitoring model that autonomously assesses breathing-type debonds between the base plate and stiffener in lightweight composite structures. The approach utilizes a specifically designed deep learning architecture that employs nonlinear ultrasonic signals for automatic debond assessment. To achieve this, a series of laboratory experiments were conducted on multiple composite panels with and without base plate-stiffener debonds. A network of piezoelectric transducers (actuators/sensors) was used to collect time-domain guided wave signals from the composite structures. These signals, representing nonlinear signatures such as higher harmonics, were separated from the raw signals and transformed into time-frequency scalograms using continuous wavelet transforms. A convolutional neural network-based deep learning architecture was designed to extract discrete image features automatically, enabling the characterization of composite structures under healthy and variable breathing-debond conditions. The proposed deep learning-assisted health monitoring model exhibits promising potential for autonomous inspection with high accuracy in complex structures that experience breathing-debonds.