The state of charge (SOC) is a parameter to describe the remaining charge of lithium-ion batteries in electric vehicles. It is a key problem to be solved in the field of electric vehicles. In this paper, ant colony optimization (ACO) algorithm is creatively applied to improve Elman neural network to form ACO-Elman neural network model, and it is applied to lithium-ion battery SOC prediction for the first time. The ACO-Elman model is trained and tested under Dynamic Stress Test and Federal Urban Driving Schedule drive profiles. The SOC estimation results of ACO-Elman model are evaluated from three aspects: mean absolute error, root mean square error, and SOC error. The results show that the ACO-Elman model has high accuracy and robustness. It has a good application prospect.

充电状态（SOC）是描述电动汽车中锂离子电池剩余电量的参数。这是电动汽车领域要解决的关键问题。本文创造性地将蚁群优化算法应用于改进Elman神经网络，形成ACO-Elman神经网络模型，并将其首次应用于锂离子电池SOC预测。ACO-Elman模型是在动态压力测试和联邦城市驾驶计划驾驶档案中进行培训和测试的。从三个方面评估了ACO-Elman模型的SOC估计结果：平均绝对误差，均方根误差和SOC误差。结果表明，ACO-Elman模型具有较高的准确性和鲁棒性。具有良好的应用前景。