Array infrared thermography for visualization of defects in bonded fiber reinforced polymer joints

Abstract

Fiber reinforced polymer (FRP) is widely used in new and existing structures, however, interfacial defects in the bonded joints pose a significant threat to structural integrity. Therefore, detection of interfacial defects is imperative for ensuring structural safety. This study proposes array infrared thermography (IRT) as a novel non-destructive evaluation method to visualize interfacial defects. Array IRT provides uniform heat excitation within the spatial domain, which overcomes the problem of heat concentration by conventional IRT. Forty-five bonded FRP plate specimens were tested using array IRT, results of which show that interfacial defects can be accurately detected within (8h+8) s (where h is the thickness of the upper layer of FRP in mm). Array IRT achieves high accuracy in detecting shapes, particularly sharp corners of defects. A pre-processing method was proposed to eliminate the twisted angles of thermal camera and to more clearly show the defects in the thermograms. A database containing tested thermograms and the corresponding predefined defects was established. Intelligent algorithms - UNet, Deeplabv3, and YOLOv8 - were used to segment the defected regions for array IRT analysis, results of which show a precision of 95.8%, 94.4%, and 94.1%, respectively.