Abstract: In this paper, a machine learning based approach is presented for detection and classification of faults in an induction machine. Five different classification algorithms, namely, support vector machine (SVM), decision tree, random forest, naive Bayes, and extreme gradient boosting (XGBoost), are adopted. The diagnosis of the most common types of faults such as broken bars, interturn fault and outer racing fault are considered. The current signatures under healthy and various faulty conditions are used for training and validating the models. The feature extraction step is implemented with the help of discrete wavelet transform (DWT). Following DWT, the features obtained are fed to the classification algorithms and subsequently the performance of each algorithm with respect to each fault condition is evaluated with appropriate metrics. Finally, a performance comparison is done and the most suitable classifier for reliable diagnosis of each of the fault condition is suggested.
Cite this article as: V. Rajini, K. B. Sundharakumar, V. S. Nagarajan, H. Karunya, H. Babu Manogaran and W. Abitha Memala, “A classification approach for induction motor faults based on empirical mode decomposition and machine learning algorithms,” Electrica, 24(2), 515-524, 2024.