基于改进粒子群优化算法和极限学习机的锂离子电池健康状态预测  

Improved particle swarm optimization and an extreme learning machine are used to predict the health state of lithium-ion batteries

作　　者：丁同 傅晓锦[1] Ding Tong;Fu Xiaojin(School of Mechanical Engineering,Shanghai Dianji University,Shanghai 201306,China)

机构地区：[1]上海电机学院机械学院,上海201306

出　　处：《国外电子测量技术》2023年第5期163-173,共11页Foreign Electronic Measurement Technology

基　　金：上海市自然科学基金(11ZR1413800)项目资助。

摘　　要：锂离子电池健康状态(SOH)的准确预测能够保障电池安全稳定的运行。针对目前SOH预测模型准确性不高的问题,提出了一种改进粒子群优化算法(IPSOVP)和极限学习机(ELM)的SOH预测模型。首先,对电池数据进行分析,选择能够映射SOH变化趋势的健康特征;然后,采用Pearson相关系数分析法选出与SOH具有极高相关性的3个健康特征作为模型的输入,SOH作为输出;利用IPSOVP算法对ELM进行优化,建立IPSOVP-ELM模型进行SOH预测;最后,利用NASA电池数据集对IPSOVP-ELM模型进行验证,并与ELM模型、PSO-ELM模型、反向神经网络(BP)以及长短期记忆网络(LSTM)模型进行比较。实验结果表明,IPSOVP-ELM模型误差稳定在2%以内,具有更高的预测精度和鲁棒性,性能更好。Accurate prediction for state of health(SOH)of lithium-ion batteries can ensure the safe and stable operation of batteries.A SOH prediction model based on improved particle swarm optimization algorithm(IPSOVP)and extreme learning machine(ELM)was proposed to solve the problem of low accuracy of the current SOH prediction model.First,the battery data is analyzed to select health features that can map the SOH trend.Then,Pearson correlation coefficient analysis was used to select three health features with high correlation with SOH as input and SOH as output.The IPSOVP algorithm was used to optimize ELM,and the IPSOVP-ELM model was established to predict SOH.Finally,the IPSOVP-ELM model is verified by using NASA battery data set,and compared with ELM model,PSO-ELM model,back propagation(BP)network and long and short term memory(LSTM)network model.The experimental results show that the error of IPSOVP-ELM model is stable within 2%,which has higher prediction accuracy,robustness and better performance.

关 键 词：锂离子电池 SOH预测 健康特征 改进粒子群算法 极限学习机 

分 类 号：TM911.18[电气工程—电力电子与电力传动]