Diagnostic and Prognostic Health Management of Electric Vehicle Powertrains : A Data Driven Approach for Induction Motor

Abstract

In recent years, the topic of electric traction has become increasingly important, with significant research efforts focused on developing more efficient and reliable electric motors for use in electric vehicles (EV). To ensure the reliable operation of these motors, it is essential to incorporate on-board diagnostic and prognostic tools that can detect and predict potential failures, induction machines have emerged as a popular choice for EV due to their durability and ability to handle high torque and variable speed applications. However, ensuring the health and reliability of these machines requires advanced diagnostic and prognostic techniques. In this context, this paper presents a study on the diagnostic and prognostic capabilities of machine learning models for an induction motor used in electric vehicle powertrain applications. The study aims to identify the most significant features of vibration signals collected from accelerometers attached to the motor, and to evaluate the effectiveness of Decision forest and decision tree machine learning algorithms in diagnosing and predicting the health of the motor. The models are trained on the full extracted features and the selected features using PCA and CA to classify the health condition of the motor. The comparison of the results from the different approaches demonstrates that the combination of the selected features using PCA with the Decision Forest algorithm achieves the best classification performance for the simulated motor fault conditions.