摘要
提出适用于变电站铅酸电池的寿命预测模型。引入集合经验模态分解法对长短期记忆神经网络模型进行改进,将电池使用寿命的特征量进行分解,构建多层结构的预测模型,提升预测结果的准确率。对比仿真结果发现,提出的改进方法可适用于小样本下的预测模型训练,且在各个样本中的平均绝对误差(MAE)不超过3 Ah,均方根误差(RMSE)不超过4 Ah。
A life prediction model for substation lead-acid battery was proposed.The ensemble empirical mode decomposition method were introduced to improve the long short-term memory(LSTM)neural network model,the characteristic quantities of service life of the battery was decomposed,a multi-layer structure prediction model was established,the accuracy of the prediction results was improved.According to the comparison of simulation results,it was found that the improved method could be applied to the prediction model training under small samples,the mean absolute error(MAE)in each sample did not exceed 3 Ah,the root mean square error(RMSE)did not exceed 4 Ah.
作者
李瑞津
刘斌
张学敏
舒征宇
LI Rui-jin;LIU Bin;ZHANG Xue-min;SHU Zheng-yu(Yuxi Power Supply Bureau of Yunnan Power Grid Limited Liability Company,Yuxi,Yunnan 653100,China;College of Electrical Engineering and New Energy,Three Gorges University,Yichang,Hubei 443000,China)
出处
《电池》
CAS
CSCD
北大核心
2020年第6期560-564,共5页
Battery Bimonthly
基金
国家自然科学基金(61876097)。