基于土地使用性质的饱和空间负荷预测研究  

Research on Saturated Spatial Load Forecasting Based on Land Use Property

作　　者：张纪伟 刘晓明 冯人海 贡卓 曹建梅 张峰 吴元香 龙剑桥 ZHANG Jiwei;LIU Xiaoming;FENG Renhai;GONG Zhuo;CAO Jianmei;ZHANG Feng;WU Yuanxiang;LONG Jianqiao(State Grid Tibet Electric Power Co.,Ltd.,Lhasa,Tibet 850000,China;State Grid Shandong Electric Power Company,Jinan,Shandong 250000,China;School of Electrical and Information Engineering,Tianjin University,Tianjin 300072,China)

机构地区：[1]国网西藏电力有限公司,西藏拉萨850000 [2]国网山东省电力公司,山东济南250000 [3]天津大学电气自动化与信息工程学院,天津300072

出　　处：《广东电力》2019年第8期35-42,共8页Guangdong Electric Power

基　　金：国家自然科学基金项目(61601309);国网西藏电力有限公司科技项目(SGXZRG00FJJS1800123);深圳市科技计划项目(JCYJ2016042293217170)

摘　　要：精准预测电力负荷对提升电网规划和建设质量意义重大,传统饱和负荷预测方法受历史信息影响较大,且预测结果的计算复杂、精度低,为此提出一种基于误差模型变换的饱和负荷预测方法。该方法首先考虑城市用地性质及开发时间,将片区地块分成同质地块和同时地块;继而提出迭代式负荷预测系统架构,将片区负荷预测问题转换为网格化分区的参数估计和模型的整合预测2个子问题相互迭代;最后利用误差模型变换改进极大似然估计,使负荷预测结果不再依赖于地块历史信息。仿真结果表明,该方法的负荷预测结果精确度较高,且受误差影响较小。Accurate forecast of power load is of great significance to promote power grid planning and construction quality.But traditional saturated load forecasting methods are greatly affected by historical information,which may cause complicated calculation on forecasting results and low precision.Therefore,this paper presents a kind of saturated load forecasting method based on error model transformation.This method firstly considers urban land use property and development time,and divides district plots into plots with the same quality and plots with the same development time.Then it proposes an iterative load forecasting system framework and transfers the district load forecasting problem into a problem of mutual iteration of two sub-problems including parameter estimation of grid partitioning and model integration forecasting.Finally,it makes use of the error model to transform the maximum likelihood estimation so as to make load forecasting results no longer dependent on historical information of plots.The simulation result indicates that accuracy of load forecasting results of this method is higher and is less affected by errors.

关 键 词：配电网 负荷预测 迭代算法 误差模型变换 极大似然估计 

分 类 号：TM715.1[电气工程—电力系统及自动化]