全国电线结冰厚度分布及等级预报模型

Characteristic Analysis of Ice Accumulation on Transmission Lines and Simulation Based on ANN Model over China

利用1961—2008年全国电线结冰及相关气象要素台站观测资料,分析电线结冰厚度的空间分布特征,并建立利用前期冰冻日数,前1天日最低气温、相对湿度、风速和降水量预报电线结冰厚度等级的3层人工神经网络BP模型。结果表明:电线结冰厚度最大值在10 mm以上的地区,在北方主要位于东北东南部、内蒙古东北部、华北中部以及甘肃南部等地,在南方主要位于安徽东南部、江西北部、湖南南部、湖北西部、重庆南部、贵州中部以及四川东部等地,呈东—西向带状分布。近48年来,安徽黄山和江西庐山覆冰厚度历年极大值和覆冰日数均呈增长趋势;四川峨眉山、甘肃西峰镇的历年极大值和覆冰日数均呈减小趋势。建立的人工神经网络BP模型能在一定程度上预报结冰厚度等级,模型对近10年的回报结果显示,准确率为81.3%。

An extreme frozen ice and snow disaster brought severe losses in early 2008 to China.A large number of overhead transmission lines are destroyed seriously during this event.It is critical to investigate the characteristics of ice accumulation and develop appropriate models to estimate transmission line icing in China.Using the observation data of ice accumulation on transmission lines and meteorological data collected from over 600 meteorological stations of China,the characteristics of ice accumulation are analyzed and an ANN model for predicting ice accumulation depth grades is developed.The results mainly include four aspects.First,the area with ice accumulation extremes over 10 mm is mainly located in the southeastern part of Northeastern China,the northeastern part of Inner Mongolia,the middle part of North China,the southern part of Gansu Province and a east—west zone to the south of the Yangtze River.Second,there are 50 stations（1/6 of total ice accumulation observation stations） experiencing extreme ice accumulation in 2008.Third,trends for yearly maximum of ice accumulation depth and ice accumulation days increase for Huangshan station in Anhui Province and Lushan station in Jiangxi Province,while decrease for Emeishan station in Sichuan Province and Xifengzhen station in Gansu Province.Finally,an ANN model based on three layer BP network is developed to predict ice accumulation thickness grades.It is used to predict ice accumulation thickness grades in recent 10 years,and the accuracy rate is 81.3% comparing with observation.For 7.5% of the samples,prediction result is one grade higher than observation and for 11.3% it is one to four grades lower.The ANN model underestimates extreme high values seriously,which suggests it needs improving.Besides,the data from weather observation stations might be different from the field surroundings.Nevertheless,in the areas with high risks of freezing hazard,the model can provide information for electricity suppliers to optimize the design of reliable equipments and to take preventive actions avoiding serious damage by ice-snow frozen weather.