A Deep Learning Approach for an Effective Speed and Torque Forecasting Policy of PMS Motors in Electric Vehicles

Abstract

Recently, with the rapid adoption of electric vehicles (EVs) for modern transportation systems, an accurate forecasting of speed and torque is an utmost priority. As permanent magnet synchronous motors (PMSM) are an integral part of such EVs, hence this work has undertaken an effective forecasting of speed and torque of such motors. To showcase the efficacy of the proposed deep learning architecture for an effective speed and torque forecasting policy, this work adopts the dataset as formulated by University of Paderborn incorporating the effects of various factors like ambient temperature, coolant temperature, stator temperature etc. Gaussian copula based synthetic data generation have been used in this paper which effectively showcases an enhancement in model performance. This work shows a critical comparison between the proposed deep learning architecture along with some machine learning models, which further promotes the efficacy of the proposed forecasting policy.