Integrating FRACAS and FMECA with Natural Language Processing (NLP): An AI-Assisted Approach to Reliability Analysis

Abstract

This study presents a novel, AI-assisted approach to industrial reliability analysis that integrates Failure Modes, Effects and Criticality Analysis (FMECA) and the Failure Reporting, Analysis, and Corrective Action System (FRACAS) with Natural Language Processing (NLP). We developed an algorithm that automates and streamlines the analysis of equipment field-failure reports and other unstructured maintenance records. The proposed framework combines unsupervised clustering to identify recurring equipment failure modes with a supervised Support Vector Machine (SVM) classifier with a Radial Basis Function (RBF) kernel to categorize equipment field reliability reports by failure mode and underlying mechanism at scale. Using an  train–test split, the proposed model achieved  accuracy on the test dataset, indicating effective generalization to unseen maintenance reports. The Confusion Matrix metrics across all classes showed true positive rate (TPR) (or sensitivity) of 0.91, indicating the model’s strong ability to correctly identify positive samples. The false positive rate (FPR) averaged 0.03 across all classes, demonstrating excellent specificity (true negative rate, TNR of 0.97). Operationally, the methodology reduces the resource-intensive manual work required to prepare, interpret, and process FRACAS reports, thus enabling timelier, data-driven equipment reliability analysis. Overall, the study demonstrates the feasibility and benefits of using AI-assisted reliability tools that balance automation with human expertise through a human-in-the-loop approach.