Application of AI to Predict PMSM Temperature

Technology transformation in mobility solution has given electric motors higher attentions. So, it's essential to understand electric motor's thermal behavior to avoid failures and improve cycle efficiency. Its cumbersome to estimate inner components temperature with available testing & simulation methods. In this work, attempt was made to analyze the electric motor sensor data at various load conditions and build a correlation matrix of various parameters. This enabled a good understanding of dependent parameters to predict the rotor and stator temperature. Critical parameters in the data set were segregated and different regression models were investigated. The outcome of Machine Learning models was not satisfactory in terms of accuracy. Hence various Deep Learning models such as ANN, CNN and RNN were considered for further evaluation. Deep Learning Models with hyper parameter tuning technique yielded 95% regression score.