Artificial intelligence based thermal management of the electric motor drive for off-road vehicles

Abstract

The advancement in the microcontroller technology with time, have improved its processing power along with its power efficiency, in addition to enhanced memory and communication capabilities. These capabilities, opens the pathway for the integration and usage of Artificial Intelligence (AI) into the embedded systems, which enables its usage into the real-time applications of automotive Electric Vehicles (EVs). Accordingly, this thesis work highlights the behavior of enlisted Machine Learning (ML) algorithms, which when applied to the target vehicle i.e., Off-Road Vehicles such as electric tractors, to achieve application requirement needs. Automotive industry is taking efforts to migrate towards EV, taking a step towards sustainability. Electric motor drives play a key role in the architecture of EV. With this importance of electric motor drives, need arises in terms of its safe operation during the lifecycle of the vehicle. Different vehicle protection measures are to be employed to prevent its failure due to thermal stress i.e., motor temperature. Detection of abnormalities in terms of rise in temperature above warning or critical motor temperature, shall allow the longetivity of the motor and lead to vehicle performance under stress conditions both physically and internally. To achieve the same, ML models were identified after doing literature study and trained on available bench mark data and then applied to actual target vehicle. Two ML algorithms i.e., Extreme gradient boosting (XGBoost) and Random Forest Regressor (RFR) along with two Deep-Learning (DL) i.e., Long Short-Term Memory (LSTM) and Convolutional Neural Network (CNN) algorithms are considered in this thesis work, to understand algorithms’ behaviour and evaluate algorithms’ performance in both ML and DL based models, when trained with real target vehicle datasets recorded from electric motor drives used in off-road vehicles.