基于VMD-SSA-LSTM的架空输电导线覆冰预测模型  

Ice-Cover Prediction Model of Overhead Transmission Conductor Based on VMD-SSA-LSTM

作　　者：陈彬[1,2] 徐志明[2] 贾燕峰 丁锐鑫 张少峰 李飚 王佳琳 CHEN Bin;XU Zhiming;JIA Yanfeng;DING Ruixin;ZHANG Shaofeng;LI Biao;WANG Jialin(Hubei Provincial Engineering Technology Research Center for Power Transmission Line(China Three Gorges University),Yichang 443002,China;College of Electrical Engineering and New Energy,China Three Gorges University,Yichang 443002,China;Sanmenxia Electric Company,State Grid Henan Electric Power Company,Sanmenxia 472000,China;Lushi Electric Company,State Grid Henan Electric Power Company,Sanmenxia 472200,China)

机构地区：[1]湖北省输电线路工程技术研究中心(三峡大学),湖北宜昌443002 [2]三峡大学电气与新能源学院,湖北宜昌443002 [3]国家电网河南省电力公司三门峡供电公司,河南三门峡472000 [4]国家电网河南省电力公司卢氏县供电公司,河南三门峡472200

出　　处：《三峡大学学报（自然科学版）》2024年第4期105-112,共8页Journal of China Three Gorges University：Natural Sciences

基　　金：国家自然科学基金项目(52107006);国网河南省电力公司科技项目(5217I0230001)。

摘　　要：针对输电导线覆冰过程间断性强且波动性大而导致的现有模型预测精度不高的问题,从覆冰厚度数据的时序信息和气象信息出发,提出一种基于变分模态分解(variational mode decomposition,VMD)的麻雀搜索算法(sparrow search algorithm,SSA)优化长短期记忆网络(long short-term memory network,LSTM)的覆冰组合预测模型.该方法首先使用VMD分解覆冰厚度数据,降低了原始序列的不稳定性,得到具有不同中心频率的本征模态分量;其次,采用SSA算法对LSTM中的3个参数进行寻优;最后,对模态分量分别建立LSTM预测模型,将各个模态分量的预测值叠加为覆冰厚度的总预测值.通过实例仿真,对所提预测模型进行验证.结果表明:VMD-SSA-LSTM组合模型与其他模型相比,其预测精度有进一步提高,验证了所提方法的有效性.In view of the problem of low prediction accuracy of the existing models caused by the strong intermittency and large fluctuation of the ice coating process on transmission conductor and starting from the correlation between the time series of ice coating thickness data,a sparrow search algorithm(SSA)based on variational mode decomposition(VMD)is proposed to optimize the long short-term memory network(LSTM)neural network for the combined ice coating prediction.The algorithm VMD first is adopted to decompose the data of ice thickness thus the instability of the original sequence is reduced and the eigenmode components with the different center frequencies are obtained.Secondly,three parameters of LSTM are evaluated through the algorithm SSA.Finally,The LSTM prediction models for each modal component are established,and the overall prediction of ice thickness is aggregated through the predicted values of each modal component.The examples verify the effectiveness of the proposed prediction model.The results indicate that the proposed VMD-SSA-LSTM hybrid model demonstrates the improvement in prediction accuracy further compared with other models.

关 键 词：输电导线 覆冰预测 变分模态分解 麻雀搜索算法 长短期记忆网络 

分 类 号：TM755[电气工程—电力系统及自动化]