Hybrid Electrical Circuit Model and Deep Learning-Based Core Temperature Estimation of Lithium-Ion Battery Cell

Abstract

Aiming to improve the reliability and durability of lithium-ion battery (LIB) systems, a hybrid electrical circuit model (ECM) and two-dimensional grid long short-term memory (2-D GLSTM) neural network (NN)-based cell core temperature estimation technique is proposed in this article to leverage their respective strengths. The ECM is used to estimate the total heat generation ( <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$Q$ </tex-math></inline-formula> ) inside the cell using the measured physical parameters such as voltage, current, and temperature. Furthermore, the value of <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$Q$ </tex-math></inline-formula> and other external measurements are used by the 2-D GLSTM NN (deep learning (DL) approach) to estimate the core temperature of the battery cell. An experimentally validated Kalman filter (KF) and equivalent electrothermal model (EETM) are used to generate a large quantity of training and testing data, under a wide range of dynamic loading and ambient temperature to validate the proposed concept. The results showed that the proposed technique can achieve root-mean-squared-error of less than 1% with unknown test data which is quite satisfactory for practical applications. A comparative study with the state-of-the-art techniques in core temperature estimation showed that the proposed scheme performed better. The highly accurate prediction results under a comprehensive operating condition confirm the reliability and generalization ability with solid robustness to the external uncertainties.