基于SAO-LSTM的输电线路微气象预测  

Micrometeorological Forecasting of Transmission Lines Based on SAO-LSTM

作　　者：陈静[1] 王玥 Chen Jing;Wang Yue(School of Electrical and Information Engineering,Anhui University of Science and Technology,Huainan,Anhui 232001,China)

机构地区：[1]安徽理工大学电气与信息工程学院,安徽淮南232001

出　　处：《黑龙江工业学院学报（综合版）》2024年第6期85-92,共8页Journal of Heilongjiang University of Technology(Comprehensive Edition)

基　　金：国家自然科学基金项目(项目编号:51874010);安徽省教育厅高校自然科学研究项目(项目编号:KJ2018A0087)。

摘　　要：针对输电线路微气象数据存在周期性和关联性导致预测精度低的问题,提出一种雪消融优化器优化长短期记忆神经网络(SAO-LSTM)的输电线路微气象预测模型。首先利用主成分分析法对微气象数据进行降维;然后利用LSTM算法构建预测模型;最后利用SAO算法优化LSTM模型的三个参数,提高预测精度。在对比实验中,SAO-LSTM对比LSTM在平均绝对误差降低至0.958,均方根误差降低至1.294,确定系数提高至96.30%。实验证明基于SAO-LSTM的输电线路微气象预测模型能有效提高预测精度。A snow ablation optimizer enhanced long short-term memory neural network(SAO-LSTM)model is proposed to address the low prediction accuracy caused by the periodicity and correlation of micrometeorological data in transmission lines.Firstly,principal component analysis(PCA)is utilized to reduce the dimensionality of the micrometeorological data.Then,an LSTM algorithm is employed to construct the prediction model.Finally,the SAO algorithm is applied to optimize the three parameters of the LSTM model to improve prediction accuracy.In the comparative experiments,the mean absolute error of SAO-LSTM compared with LSTM is reduced to 0.958,the root mean square error is reduced to 1.294,and the determination coefficient is increased to 96.30%.The results demonstrate that the transmission line micrometeorological prediction model based on SAO-LSTM can effectively improve prediction accuracy.

关 键 词：输电线路 微气象预测 雪消融优化器 LSTM神经网络 

分 类 号：TM726.3[电气工程—电力系统及自动化] TP18[自动化与计算机技术—控制理论与控制工程]