Core Temperature Estimation Method of Lithium-ion Battery Based on An Improved Recurrent Neural Network Algorithm

Estimating the core temperature of the battery is not only an important task of the battery management system, but also a key means to ensure the safety and reliability of the battery for other important purposes. The core temperature of the battery cannot be observed directly, but it has a certain correlation with other battery measurable parameters (current, voltage, ambient temperature, surface temperature). Therefore, this paper proposes an iteration method based on an improved recurrent neural network algorithm (RNN) whose name is simple recurrent units (SRU) to calculate this correlation, and then estimates the core temperature of the battery under high dynamic driving condition. The experimental results show that this method can accurately estimate the temperature of lithium-ion battery cell of electric vehicle under wide ambient temperature, and mean square error is than 0.604566.