Analysis of the evolution of internal corrosion cracks in RMOCC compared with GLCM and SVM

Abstract

In view of the difficulty of identifying internal micro corrosion-induced cracks in concrete and the poor accuracy of quantitative analysis that results in inaccurate results of the formation law of internal cracks, RMOCC was subjected to a galvanostatic acceleration test, and X-CT technology was combined with the Support Vector Machines (SVM) identification algorithm and Grey-Level Co-Occurrence Matrix (GLCM) theory. Using the SVM algorithm and GLCM theory, the internal average crack width of concrete and the contrast, correlation, angular second moment (ASM), and inverse difference moment (IDM), which characterize the change in slice texture information, were used as degradation parameters, respectively. Using the average internal crack width and IDM as the degradation index, a reliability degradation competition failure analysis was conducted to study RMOCC's internal crack formation law. The results showed that the SVM algorithm had a greater than 95% accuracy in recognizing cracks. In the entire corrosion-induced crack formation process, IDM and the average internal crack width values were consistent with the normal distribution. Through reliability degradation competition failure analysis between IDM and the average crack width value, the average crack width calculated with SVM is more suitable for the degradation analysis of internal corrosion-induced cracks in RMOCC.