GaussianNB-Based Anomaly Detection Method for Automotive Motor Bearing

In this paper, an unsupervised anomaly detection model based on GaussianNB is proposed to identify abnormal conditions in the motor bearing system of electric vehicles. Statistical features (such as the mean value and root mean square) are extracted from vibration signals to construct a feature matrix. Additionally, the angular deviation between samples is computed using the GaussianNB model to enable accurate detection of anomalies in unlabeled data. Experimental results demonstrate that the proposed model excels in terms of false alarm rate and accuracy, effectively distinguishing between normal and abnormal samples. This validates the effectiveness and practicality of the method for anomaly detection in electric vehicle motor bearings. The model presented in this study does not depend on labeled samples. It is characterized by simple computation and high detection efficiency, offering a practical solution for anomaly detection and fault prediction in electric vehicle motor bearings.