基于特征选择的LightGBM算法预测钠离子电池剩余寿命  

Predicting remaining useful life of sodium-ion batteries using feature selection-based LightGBM algorithm

作　　者：史永胜[1] 翟欣然 栾飞[2] 胡玙珺 SHI Yong-sheng;ZHAI Xin-ran;LUAN Fei;HU Yu-jun(School of Electrical and Control Engineering,Shaanxi University of Science&Technology,Xi′an 710021,China;College of Mechanical and Electrical Engineering,Shaanxi University of Science&Technology,Xi′an 710021,China)

机构地区：[1]陕西科技大学电气与控制工程学院,陕西西安710021 [2]陕西科技大学机电工程学院,陕西西安710021

出　　处：《陕西科技大学学报》2024年第2期174-181,共8页Journal of Shaanxi University of Science & Technology

基　　金：国家自然科学基金项目(22279076,61871259)。

摘　　要：钠离子电池剩余使用寿命(RUL)的准确预测对于可再生能源系统中的大规模储能设备具有重要意义.提出了一种基于特征选择的LightGBM方法来预测钠离子电池的剩余使用寿命.通过结合Pearson相关系数和灰色关联度,选择了四个与电池寿命高度相关且不同特征之间自相关程度较低的最佳特征.采用LightGBM模型,并结合网格搜索(GridSearchCV)算法对其超参数进行优化,以达到最佳的预测性能.通过钠离子电池数据验证了预测模型的优越性,并与使用GridSearchCV算法的GBRD和RF模型在相同条件下进行比较.结果表明,该方法能够显著加快模型运算速度,并相比传统算法具有更高的可靠性和更好的预测性能,预测的最大MAE、MSE、RMSE分别不超过3.0、17.7、4.2.Accurate prediction of the remaining useful life(RUL)of sodium-ion batteries is of significant importance for large-scale energy storage systems in renewable energy applications.In this study,a feature selection-based LightGBM method is proposed to predict the RUL of sodium-ion batteries.By combining Pearson correlation coefficient and grey relational degree,four optimal features with high correlation to battery life and low autocorrelation between different features are selected.The LightGBM model is employed,and its hyperparameters are optimized using the grid search cross-validation(GridSearchCV)algorithm to achieve the best prediction performance.The superiority of the prediction model is validated using sodium-ion battery data and compared with the GBRD and RF models with GridSearchCV algorithm under the same conditions.The results demonstrate that this method significantly speeds up the model computation,and it exhibits higher reliability and better prediction performance compared to traditional algorithms.The maximum MAE,MSE,and RMSE of the predictions do not exceed 3.0,17.7,and 4.2,respectively.

关 键 词：钠离子电池 剩余使用寿命 LightGBM 特征选择 

分 类 号：TM912[电气工程—电力电子与电力传动]