Regression Analysis in Electrical Engineering Applications: A Machine Learning Approach

Abstract

In the context of the Battery Management System (BMS) of an Electric Vehicle (EV), accurately predicting the terminal voltage of the battery is of utmost importance. However, this prediction model is dependent on the battery’s chemistry and its overall lifespan. To address this issue, this work presents a generalized procedure for implementing a Machine Learning (ML) based prediction model. Specifically, we compare the performance of five distinct regression techniques, namely, decision tree, ensemble boost and bagg, support vector machine, and neural network, using a supervised ML approach. The performance of the different regression techniques is evaluated by means of the Root Mean Square Error (RMSE). The proposed method of using ML techniques to develop an accurate prediction model for a specific task, as discussed in this work, has the potential to be implemented in various other regression tasks of engineering applications. Therefore, the approach presented in this work can serve as a blueprint for developing accurate prediction models in other engineering applications, provided that the relevant data and training are available.