A Study on the Statistical Analysis and Artificial Intelligence Model to Improve the Reliability of Safety Inspection

Abstract

In this study, safety diagnosis results from underground and nuclear power plant structures were collected to evaluate the reliability of on-site structural safety assessments. The analysis of these results revealed a discrepancy between the compressive strength determined by rebound hardness and the core compressive strength, with the former typically being evaluated higher than the latter. Additionally, existing strength prediction models did not adequately explain field data, whereas artificial intelligence models, particularly the support vector machine model, demonstrated improved accuracy and reduced error rates. This indicated the superior performance of support vector machine models in this context.