基于PSO寻优与DBN神经网络的电晕损耗预测  

Corona Loss Prediction of UHV AC Transmission Line Based on DBN Neural Network Optimized by PSO

作　　者：黄书民 蒋林高 李志川 杨光绪 宋福根[2] HUANG Shumin;JIANG Lingao;LI Zhichuan;YANG Guangxu;SONG Fugen(Ultra High Voltage Branch Company of State Grid Fujian Electric Power Co.,Ltd.,Fuzhou 350013,China;College of Electrical Engineering and Automation,Fuzhou University,Fuzhou 350108,China)

机构地区：[1]国网福建省电力有限公司超高压分公司,福建福州350013 [2]福州大学电气工程与自动化学院,福建福州350108

出　　处：《中国电力》2022年第6期95-102,214,共9页Electric Power

基　　金：国家电网有限公司科技项目(52130A200005)。

摘　　要：特高压交流输电线路的电晕损耗与降雨量、比湿、温度、相对湿度、压强等天气条件有相关性,可通过部分天气条件对特高压交流输电线路电晕损耗进行预测,提出了一种特高压交流输电线路的电晕损耗预测方法。根据粒子群优化算法(particle swarm optimization,PSO)寻优机制与深度信念网络(deep belief network,DBN)预测原理,详细说明了该预测方法的智能算法机制,并提出了一套完整的基于PSO—DBN智能算法的预测方法。首先,通过斯皮尔曼相关系数的大小确定与电晕损耗有较强相关性的天气条件,并作为特征值;然后以所选特征值为指标体系构建DBN神经网络进行电晕损耗预测,再采用PSO寻优算法对DBN神经网络进行内部参数调整,提升DBN神经网络的预测准确性;最后利用所提算法对实际运行的闽浙特高压输电线路的电晕损耗进行算法预测,与该线路的运行统计电晕损耗值进行对比分析,验证了所提预测方法的可行性。该方法为特高压输电线路电晕损耗研究和工程设计提供参考。Based on the correlation between corona loss of UHV AC transmission lines and weather conditions such as rainfall,specific humidity,temperature,relative humidity,pressure,etc.,a corona loss prediction method for UHV AC transmission lines is proposed by predicting corona loss of UHV AC transmission lines under some weather conditions.Based on the optimization mechanism of particle swarm optimization(PSO)and the prediction principle of deep belief network(DBN),the intelligent algorithm mechanism of this prediction method is explained in detail,and a complete set of prediction methods based on PSO—DBN intelligent algorithm is proposed.Firstly,the weather conditions that have strong correlation with corona loss are determined by the size of the Spearman correlation coefficient and used as the characteristic values.Then the DBN neural network is built with the selected eigenvalues as the index system to predict the corona loss.Then the internal parameters of the DBN neural network are adjusted by PSO optimization algorithm to improve the prediction accuracy of the DBN neural network.Finally,the proposed algorithm is used to predict the corona loss of the actual Fujian—Zhejiang UHV transmission line,and the statistical corona loss values of the line are compared and analyzed to verify the feasibility of the proposed prediction method.This method provides reference for corona loss research and engineering design of extra high voltage transmission lines.

关 键 词：特高压 电晕损耗 斯皮尔曼相关系数 粒子群优化算法 DBN神经网络 

分 类 号：TM75[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]