Evaluation of One-Class Classifiers for Fault Detection: Mahalanobis Classifiers and the Mahalanobis–Taguchi System

Today, real-time fault detection and predictive maintenance based on sensor data are actively introduced in various areas such as manufacturing, aircraft, power system monitoring. Many faults motors or rotating machinery like industrial robots, aircraft engines, wind turbines can be diagnosed by analyzing signal vibration noise. In this study, to detect failures data, preprocessing was performed using processing techniques the Hamming window cepstrum transform. After that, 10 statistical condition indicators were extracted train machine learning models. Specifically, two types of Mahalanobis distance (MD)-based one-class classification methods, MD classifier Mahalanobis–Taguchi system, evaluated detecting machinery. Their performance for with different imbalanced ratios comparing binary models, which included classical versions support vector random forest algorithms. The experimental results showed MD-based classifiers became more effective than cases there much fewer defect normal is often common real-world field.