Detecting Faults in Electric Motors Based on Current and Rotor Speed Measurement Using a Naive Bayesian Classifier

Automatic fault detection in electric motors is essential for maintaining reliability and reducing downtime in industrial applications. This paper presents a novel approach that applies a Gaussian Naive Bayes classifier to high-frequency measurements of current and rotor speed to identify motor faults. Data were collected from a dedicated test bench at AGH University of Krakow, where motors were evaluated under healthy conditions as well as under mechanical, electrical, and combined damage scenarios. Statistical features were extracted from the time-series signals and used to train the classifier. Experimental results indicate that the classifier achieved detection accuracies of approximately 76 percent using current data and 85 percent using rotor speed data, with similar performance when both modalities were combined. These findings demonstrate the potential of probabilistic models for real-time fault diagnostics and lay the groundwork for future research into advanced signal fusion and classification techniques.