Gated Recurrent Units-Assisted State-Space Modeling for Electric Vehicle Temperature Prediction

Abstract:
The electric system (electric motor, electric control, and battery pack) is the most critical component of e-mobility. Effective thermal management of the electric system is crucial for enhancing the range, reliability, and performance of e-mobility. Accurate prediction of the electric system temperature forms the foundation of effective thermal management. This paper presented a novel gated state-space model (GSSM) that utilized the gated recurrent unit (GRU) to assist the state-space model (SSM) in predicting electric system temperature. The gated state-space model is a physics-informed method, while it offers consistency with the efficiency of the data-driven method and dynamics system properties. The proposed method is experimentally verified with a real electric system temperature dataset provided by an electric vehicle (EV) company.