AI-Assisted Root Cause Analysis for Mechanical Failures in Nuclear Power Plants

Abstract

This project presents the development and evaluation of an artificial intelligence system designed to improve the accuracy and efficiency of diagnosing mechanical failures in turbine systems within nuclear power plants. Traditional root cause analysis relies heavily on manual expertise, which leads to extended diagnostic times, inconsistent classifications, and potential human error. The methodology used followed the DMAIC process improvement framework to integrate supervised machine learning techniques using failure data. A simulation was implemented to test the system's ability to detect faults like bearing damage, rotor imbalance, and lubrication issues. The model was able to achieve all three objectives with the following results: an accuracy of 87.67%, reduced diagnostic time by 30%, and consistent classification across three failure categories. This confirms that AI can significantly support safer, faster, and more consistent nuclear power plant's turbine diagnostics.