神经网络在配网电能损耗预测中的应用研究  

Application of Neural Network in Prediction of Power Loss in Distribution Network

作　　者：王昕 曹敏[1,2] 邢士发 李英娜[3] WANG Xin;CAO Min;XING Shi-fa;LI Ying-na(Yunnan Power Grid Co., Ltd. Electric Power Research Institute, Kunming 650217, China;unnan Electric Power Research Institute ,Kunming 650217,China;Kunming science and technology light Testing Technology Co., Ltd. ,Kunming 650051,China)

机构地区：[1]云南电网有限责任公司电力科学研究院,云南昆明650217 [2]中国南方电网公司电能计量重点实验室,云南昆明650217 [3]昆明理工光智检测科技有限公司,云南昆明650051

出　　处：《软件》2017年第8期210-214,共5页Software

基　　金：国家自然科学基金(51567013);国家自然科学基金(51667011)

摘　　要：配网馈线和节点较多、负荷的多变,使得配网中的技术线损和管理线损分析预测困难。本文结合等值电阻法理论线损计算模型、BP神经网络、广义回归神经网络GRNN模型,对配电电网线损、电能损耗进行分析与计算,将理论应用到区域实测负荷数据和电量数据的分析中,分析了S区域线损率、线损和电能损耗,计算出了配电网电能损耗各元件所占的百分比。测试结果表明,BP模型对线损预测的均方误差为2.71;并在此基础上,考虑配电网变压器等损耗,利用PSO-GRNN模型对配电网的电能损耗进行预测,配网电能总损耗预测的均方误差为0.36,为区域电能损耗分析和降损工作提供了关键状态参数。There are many feeders and nodes in the power distribution network,so that the technical line loss and management line loss analysis in the distribution network are difficult to predict.In this paper,By Combining with the theoretical line loss calculation model,BP neural network,PSO-GRNN model The line loss and power loss of the distribution grid are analyzed and calculated.The theory is applied to the analysis of the measured load data and the electricity data.The loss rate,line loss and power loss of the S region are analyzed,and the percentage of each component of the electrical energy loss in the distribution network is calculated.The results show that the mean square error of the BP model is2.71.On this basis,the PSO-GRNN model is used to predict the power loss of the distribution network.The mean square error is0.36,which is the regional power loss analysis and Loss of work provides critical state parameters.

关 键 词：配网线损计算 BP神经网络 PSO-GRNN算法 线损预测 

分 类 号：TM73[电气工程—电力系统及自动化]