Spatiotemporal Modeling of Lithium Battery Temperature Field Based on Spectral Method and Broad Learning System

Effective temperature prediction is critical for the property and security of lithium-ion batteries. However, because the temperature field of lithium battery has the characteristics of spatiotemporal coupling and strong nonlinearity, it is very difficult to model. Thus, this paper proposes a spatiotemporal modeling scheme for the thermal process of lithium batteries based on spectral method and broad learning system (BLS). According to the mechanism model of thermal process, the dimension of thermal process is reduced by time-space (T/S) separation method. Firstly, a group of suitable spatial basis functions are obtained according to the spectral method to process the spatial dimension information, and then the corresponding low-order time coefficients are obtained by T/S separation. Finally, the time coefficients are modeled and trained using BLS. The experimental research proves the effectiveness of this modeling method.