一种基于历史数据的输电线路覆冰增长快速预测方法  

A Rapid Prediction Method for Icing on Transmission Lines Based on Historical Data

作　　者：李昭廷[1] 郝艳捧[1] 

机构地区：[1]华南理工大学电力学院,广州510640

出　　处：《电瓷避雷器》2012年第1期1-7,共7页Insulators and Surge Arresters

基　　金：国家重点基础研究发展计划项目(973)(编号:2009CB724507);国家高技术研究发展计划(863计划);"十一五"国家科技支撑计划项目(编号:2009BAA23B02);中央高校基本科研业务费专项资金项目(编号:2009ZZ0008)

摘　　要：2008年初极端冰雪灾害给中国南方输电线路造成了极大破坏,引起了对架空输电线路覆冰模型研究的重视。导线覆冰增长的物理过程极为复杂,受气候环境多变和气象、季节、地形地理条件、海拔、线路走向、导线本身等各种复杂因素的影响,给基于在线监测数据预测短期覆冰增长带来困难。提出一种基于在线监测的覆冰厚度历史数据的覆冰发展趋势快速预测方法——组合灰色神经网络预测模型。首先远程采集力学数据,根据模型求得覆冰厚度的历史数据;然后分别建立GM(1,1)、Verhulst和DGM(1,1)三种灰色模型,得到三组覆冰厚度趋势数据;最后将三组数据作为神经网络输入,得到覆冰趋势曲线。以云南电网和广西电网典型的两次不同类型覆冰过程为例进行了验证。结果表明,在缺乏微气象和地形条件的贫信息状态下,此模型在覆冰增长快速预测中是适用的,有效的。Extreme ice rain in early 2008 destroyed struck the power system infrastructures in Southern China,which has called lots of attentions on modeling of icing on overhead transmission lines.Due to complex physical process of icing conductors,which was affected by the various complex factors such as changeable climate and the weather,the season,the terrain,the altitude,the alignment and the conductor itself,etc,it was difficult to predict quickly short-term icing by the data from online monitoring.A way to build a combined gray neural network model was constructed to predict the trend of icing,based on historical data through remote online monitoring system.At first,mechanical data was collected online and historical data about icing thickness was calculated.Three gray models were built to get three sets of data about the trend of icing thickness and three sets of data were as input of neural network to get the curve of icing trend.Two typical different types of ice process in Guangxi Power Grid and Yunnan Power Grid were taken as examples.The results showed that the model is applicable and effective for the rapid prediction for icing thickness.

关 键 词：覆冰 输电线路 灰色预测 组合灰色神经网络 在线监测 历史数据 

分 类 号：TM75[电气工程—电力系统及自动化]