Thermal-statistical approach for diagnosis of bus bar degradation in MCFC power plant

This study presents an innovative thermal-statistical approach for diagnosing the condition of nickel-coated copper bus bars in MCFC power plant. The primary objective was to develop a non-invasive condition monitoring solution for busbars that are typically difficult to access. The method was validated using a substantial sample size of 108. The research revealed that significant changes in bus bar condition occur at temperatures exceeding 400 °C. Moreover, it was observed that the relevance of independent variables increases markedly with current flow. To optimize the diagnostic process, the study compared three different models: logistic regression, SVM, and KNN. Among these models, logistic regression demonstrated the highest overall accuracy at 93.5 %, with a specificity of 89.3 %. SVM showed comparable performance, achieving the highest specificity of 90.9 % and an accuracy of 90.7 %. In contrast, KNN exhibited relatively lower performance with 85.2 % accuracy and 80.0 % specificity. The findings confirmed that irreversible losses during current flow transform into thermal energy, leading to busbar deterioration. This thermal-statistical approach offers a novel methodology for effectively monitoring system conditions in challenging environments, potentially enhancing the reliability and efficiency of industrial applications where direct access is limited. • Statistical approaches should be taken in uncertain situations. •State of Ni-coated copper bus bar changes when it is over 400 °C. •Five significant variables achieve a specificity of 89.3 %. •Improving Classification by Optimizing Independent Variables.