基于特征多项式与改进鲸鱼算法的电池SOH预测  

Battery SOH Prediction Based on Features Polynomial and Improved Whale Algorithm

作　　者：闫羲 赖强[2] 戴晓强[1] 李奇 鄢然 YAN Xi;LAI Qiang;DAI Xiaoqiang;LI Qi;YAN Ran(College of Automation,Jiangsu University of Science,Zhenjiang 212100,Jiangsu,China;Shanghai Marine Equipment Research Institute,Shanghai 200031,China;Hudong Heavy Machinery Co.,Ltd.,Zhenjiang 212000,Jiangsu,China)

机构地区：[1]江苏科技大学自动化学院,江苏镇江212100 [2]上海船舶设备研究所,上海200031 [3]沪东重机有限公司,江苏镇江212000

出　　处：《船舶工程》2024年第11期105-112,共8页Ship Engineering

摘　　要：锂电池作为新能源船舶系统的核心设备,对其健康状态(SOH)进行准确预测有利于系统能量管理和船舶安全运行。为提高电池SOH的预测精度,提出一种多健康特征(MHF)融合和改进鲸鱼优化算法(IWOA)相结合的预测方法。在传统支持向量回归作为预测方法的基础上,通过皮尔逊分析法选取4个典型健康特征(HF),采用加权方法构建融合多个HF的多项式模型。考虑到特征的权值系数和SVR的惩罚系数C、核参数δ以及最大误差ε的取值对预测精度的影响,使用IWOA对模型中的权值系数以及3个超参数进行联合寻优。仿真结果表明,所提出的MHF-IWOA-SVR方法具有更高的预测精度与拟合度,预测误差基本保持在±0.5%以内。Lithium battery is the core equipment of new energy ship system,and accurate prediction of its state of health(SOH)is conducive to system energy management and safe operation of the ship.In order to improve the prediction accuracy of battery SOH,a prediction method combining multiple health features(MHF)fusion and improved whale optimization algorithm(IWOA)is proposed.Based on the traditional support vector regression(SVR)as the prediction method,four typical health features(HF)are selected by the Pearson analysis method,and a polynomial model is constructed by using the weighted method to fuse multiple HF.Considering the influence of the weight coefficient of the feature and the penalty coefficient C of SVR,the kernel parameterδand the maximum errorεon the prediction accuracy,the IWOA is used to jointly optimize the weight coefficient and three hyperparameters in the model.The simulation results show that the proposed MHF-IWOA-SVR method has higher prediction accuracy and better fit,and the prediction error is basically kept within±0.5%.

关 键 词：健康特征(HF) 支持向量回归 改进鲸鱼优化算法(IWOA) 健康状态(SOH) 

分 类 号：TM912[电气工程—电力电子与电力传动]