A Multi-Module Fusion Augmented Neural Network for Lithium-Ion Battery State of Charge Estimation

Accurately estimating the state of charge (SOC) of lithium-ion batteries is a core technical aspect of improving the efficiency of energy management systems and ensuring the safe and reliable operation of batteries. Traditional machine learning-based methods for SOC estimation in lithium-ion batteries suffer from limited estimation accuracy and insufficient generalization capability. To address this challenge, this paper proposes an enhanced hybrid neural network model for high-accuracy SOC prediction. Specifically, this paper proposes a deep neural network prediction model based on a multi-module structure. Its main contribution lies in the fusion of a bi-directional temporal convolutional network (BiTCN) and long short-term memory neural network (LSTM) to construct a multi-modular prediction model that effectively combines the complementary advantages of both. Meanwhile, the introduction of the self-attention (SA) mechanism enhances the ability to focus on features. Finally, hyperparameter tuning is performed by the immune particle swarm optimization algorithm (IPSO). Multi-module hybrid neural network models exhibit higher prediction accuracy. The experimental results fully validate the effectiveness of the proposed method, and the model shows excellent prediction accuracy and generalization under different battery types, different temperatures, and two operating conditions.