基于多源异构气象数据融合的线路状态识别算法  

Line state recognition algorithm based on multi⁃source heterogeneous meteorological data fusion

作　　者：于玉宗 刘玉鹏 汲倩倩 YU Yuzong;LIU Yupeng;JI Qianqian(Software Development Center,CHN Energy Network Information Technology Co.,Ltd.,Beijing 100011,China)

机构地区：[1]国能网信科技(北京)有限公司软件开发中心,北京100011

出　　处：《电子设计工程》2024年第12期157-161,共5页Electronic Design Engineering

基　　金：国网冀北电力有限公司电力科学研究院(B1018K21000B)。

摘　　要：针对传统输电线路状态判断方法存在费时费力且准确率偏低的问题,文中在多源异构气象数据融合分析的基础上提出了一种输电线路状态判别算法。该算法由气象数据分析模块和输电线路故障判断模块组成,对于LSTM网络中存在超参数选择困难的问题,利用WOA算法进行自动优化,以实现对气象数据的特征提取。而输电线路故障判断模块使用航拍线路图像作为训练集,通过引入GAN网络完成训练,并采用AE网络进行改进,从而提升原网络的性能。在实验测试中,所提算法在数据处理层面的MAE仅为0.1159,其余多个测试指标也均优于对比算法,同时还可根据训练得到的多源数据特征来对当前线路状态进行准确预测,表明了该算法具有良好的综合性能与工程实践能力。Aiming at the problem of time⁃consuming,laborious and low accuracy of traditional transm⁃ission line state judgment methods,a transmission line state judgment algorithm is proposed based on multi⁃source heterogeneous meteorological data fusion analysis.The algorithm is composed of meteor⁃ological data analysis module and transmission line fault judgment module.For the difficulty of super parameter selection in LSTM network,WOA algorithm is used for automatic optimization to achieve feature extraction of meteorological data.The transmission line fault diagnosis module uses aerial line images as the training set,completes the training by introducing the GAN network,and improves the performance of the original network by using the AE network.In the experimental test,the MAE of the proposed algorithm at the data processing level is only 0.1159,and other test indicators are better than the comparison algorithm.At the same time,the current line state can be accurately predicted according to the characteristics of multi⁃source data obtained from training,which shows that the algorithm has good comprehensive performance and engineering practice ability.

关 键 词：多源异构数据 长短时神经网络 鲸鱼算法 条件对抗网络 自编码器 输电线路保护 

分 类 号：TP391[自动化与计算机技术—计算机应用技术] TN911.2[自动化与计算机技术—计算机科学与技术]