Prediction and Analysis of Permanent Magnet Synchronous Motor parameters using Machine Learning Algorithms

Abstract

The widespread acceptance of PMSM as the motor of choice for electric vehicles, along with various other applications demands the need of stringent monitoring of temperature in order to avoid increased temperatures. Temperature values beyond a specific range can lead to major operational problems in Permanent Magnet Synchronous Motor(PMSM) along with additional maintenance costs. Using r2 values this paper compares the performance of three different Machine Learning Algorithms in the estimation of parameters in a Permanent Magnet Synchronous Motor. Making use of a pre existing test set, the accuracies in predictions by the following models are compared: Support Vector Regressor, Random Forest Regressor and Polynomial Regression. Random Forest Regression shows the highest r2 values(statistical way of knowing variation of dependent variables explained by independent variables for a particular regression model) which proves the accuracy of the model.