Automated Thermography Cognitive Sensing-Feedback Inspection for Large Irregular Sample

As an effective method for detecting inner defects in composite materials, optical pulsed thermography (OPT) nondestructive testing (NDT) is widely used in the aircraft industry. Due to the complex structure of the large irregular composite materials, traditional manual inspection leads to uneven heat conduction, low defect detection rate, and lack of automation. Studying the physical properties of heat conduction found that surface thermal uniformity is particularly sensitive to defect positioning errors as well as resolution. This article proposes an adaptive defect detection method based on thermal perception guided Robot-sensing-feedback controlling within the heat flux density isobaric surface (HFDIS) projection. In particular, it is constructed by the physical projection of the OPT as embedded with the 3-D model of the sample. HFDIS can be simultaneously used to reconstruct the 3-D thermography of the irregular sample, introduce closed-loop constraints, globally optimize the heat flux density uniformity loss function, and control the detection pose of the robotic arm. Thus, it improves heat conduction uniformity and defect detection accuracy and efficiency. Both simulation and experimental verification were conducted on multiple types of heterogeneous and special-shaped samples provided by the aircraft company, and the effectiveness and scientific validity of the detection method were rigorously evaluated.