Research On The Key Technology Of SiC Coating Defects Detection Using Pulsed Infrared Thermal Wave Non-destructive Testing Method

SiC coating structure has been gradually used in high temperature environments such as aviation, aerospace, internal combustion engines and nuclear reactors, due to its high temperature strength, high fracture toughness, resistance to erosion and so many excellent features. The complex reparation progress and poor service environment of SiC coating structure components, result to defects such as uneven coating thickness, coating spalled off the substrate and innerdefects in the substrate, etc., which have a di-rect impact on the operation controllability and safety of the apparatus. Infrared thermal wave nondestructive testing technology (IT NDT) has many advantages such as non-contactivity, intuition and high detection efficiency, and provides a new method for to detect defects in SiC coating structure. This article carry out a systematic and thor-ough study, including the theoretical analysis and simulation studies of pulsed infrared thermal wave detection principle, the building up of testing system, the selection of de-tection parameters, the processing algorithms of pulsed infrared image sequence, uni-formity of coating thickness detection and the Identification and judgment of de-fects geometric feature.The heat transfer process of the coating structure with optical pulse excitation has been studied, and the analytical model of surface temperature distribution has been es-tablished, which indicates the analytical relation between coating thickness and temper-ature difference. The infrared thermal wave non-destructive testing process has been simulated using the finite element method, and the relation between the defect geometry, test parameters and the surface temperature signals. The detection ability of pulsed in-frared thermal wave detection method used to inspect uneven coating thickness of SiC coated high temperature superalloy substrate, inner defects of SiC coated high tempera-ture superalloy substrate, and inner defects of SiC coated C/C composite substrate.The infrared thermal wave testing system with flash pulse excitation was built up, and used to detect defects of SiC coating structure with efficiency and reliablity. Through experimental research, the relation between the defect geometry, test parame-ters and the surface temperature signals has been analyzed, which verifed the correct-ness of the theoretical model. And the reasonable range of test parameters has been got.The processing algorithms of pulsed infrared thermography sequence, including the polynomial fitting the derivative time-related coefficient method, pulse phase thermography, and Markov-principal component analysis method, have been studied and used to extract the temperature signal’s feature. The signal to noise ratio of feature images have been improved, and the proposed Markov-PCA algorithm improved the signal to noise ratio significantly, which enhanced defects detection capability.Coating thickness detection using pulsed infrared thermal imaging method has been conducted, and the inverse heat conduction problem of coating thickness distribu-tion based on normalized surface temperature distribution has been established. The quantitative inversion of coating thickness has been achieved by simulated annealing algorithm. The inspection results of coating thickness of SiC coated high temperature superalloy substrate show that, while the coating thickness is in the range of45~130μm, the maximum coating thickness detection error is less than10%, comparing to conventional eddy current scan results.Thorough study of the geometric feature identi-fication and judgment of defects in SiC coating structure has been carried out. A fuzzy C-means clustering-Canny operator edge detection algorithm for infrared signature im-age defects edge extraction has been proposed, and used to the recognition of defects geometry feature. A BP neural network model has been built, taking the principal com-ponent characteristics parameters arrived through Markov-PCA analysisas as the input, and the radial size and depth of defects as the output. The built network model was trained by the collected sample data, and the radial dimension and depth of internal de-fects in SiC coated C/C composite has been got. The results show that the diameter and depth prediction error is in the range of about4%to10%, for internal defects with deep-diameter ratio of1.2to4.0, and depth1.0mm to2.5mm.