摘要
提出了一种基于多健康因子的改进粒子群优化算法（IPSO）优化长短期记忆（LSTM）模型的锂离子电池健康状态（SOH）估计方法。从在线测量的电压、电流、温度曲线中提取13个备选健康因子，利用皮尔逊相关系数分析最终获得4个健康因子作为IPSO-LSTM模型的输入，通过试验验证所选健康因子的适应性，实现SOH准确预测。取每个电池样本的50%，60%，70%数据作为训练集，其余作为测试集，与PSO-LSTM，LSTM方法作对比，试验结果表明，SOH估计的MAE，RMSE，MAPE均保持在1%以内，模型具有较强的泛化性及有效性。
Abstract
An improved particle swarm optimization(IPSO) algorithm based on multiple health factors was proposed to optimize the LSTM model for lithium-ion battery health state(SOH) estimation. Thirteen alternative health factors were extracted from the volt, current and temperature curves measured online. Pearson correlation coefficient analysis was used to obtain four health factors as the input of IPSO-LSTM model. The adaptability of selected health factors was verified by experiments and the accurate SOH prediction was achieved. The 50%, 60% and 70% data of each battery sample were taken as the training set, and the rest were taken as the testing set. Compared with the PSO-LSTM and LSTM methods, the test results showed that the MAE, RMSE and MAPE estimated by SOH were all kept within 1%, and the model had strong generalization and effectiveness.
关键词
锂离子电池 /
预测模型 /
健康管理
Key words
lithium-ion battery /
prediction model /
health management
李珺，陈小然，徐亮.
基于多健康因子和IPSO-LSTM模型的锂电池健康估计[J]. 车用发动机. 2025, 0(1): 39-46
LI Jun，CHEN Xiaoran，XU Liang.
Lithium Battery Health Estimation Based on Multiple Health Factors and IPSO-LSTM Model[J]. Vehicle Engine. 2025, 0(1): 39-46
{{custom_sec.title}}
{{custom_sec.title}}
{{custom_sec.content}}
