Quantitative detection of defects based on Markov-PCA-BP algorithm using pulsed infrared thermography technology

Quantitative detection of debonding defects' diameter and depth in TBCs has been carried out using pulsed infrared thermography technology. By combining principal component analysis with neural network theory, the Markov-PCA-BP algorithm was proposed. The principle and realization process of the proposed algorithm was described. In the prediction model, the principal components which can reflect most characteristics of the thermal wave signal were set as the input, and the defect depth and diameter was set as the output. The experimental data from pulsed infrared thermography tests of TBCs with flat bottom hole defects was selected as the training and testing sample. Markov-PCA-BP predictive system was arrived, based on which both the defect depth and diameter were identified accurately, which proved the effectiveness of the proposed method for quantitative detection of debonding defects in TBCs. (C) 2016 Elsevier B.V. All rights reserved.