一种规范化的处理相关因素的短期负荷预测新策略  

A NEW UNIFIED APPROACH TO SHORT-TERM LOAD FORECASTING CONSIDERING CORRELATED FACTORS

作　　者：康重庆[1] 程旭[1] 夏清[1] 沈瑜[1] 

机构地区：[1]清华大学电机系,北京100084

出　　处：《电力系统自动化》1999年第18期32-35,共4页Automation of Electric Power Systems

摘　　要：提出了一种规范化的可以直接考虑各种相关因素的短期负荷预测新策略。这种策略不仅可以考虑已经被详细研究过的气象因素，而且可以同时考虑一般性的分类指标，如工作日／休息日，正常日／节假日等。设计了一个规范化的相关因素映射数据库，用以反映不同影响因素在映射函数上的差异。根据模式识别的基本原理，应用聚类分析方法描述由于相关因素的不同而导致的待预测日与历史日之间的差异程度。作为这种预测策略的一个应用，实现了一种基于相关因素匹配的短期预测方法。实际算例表明，应用这种新的规范化的预测策略可以得到更加精确的预测结果。A new unified approach to short term load forecasting is proposed w hich is capable of dealing w ith m ost of thefactors correlated w ith pow er system load. This approach can consider notonly w eather w hich has been studied thoroughly insom e previous literatures⒙but also som e classification indexes such as w eekday/w eekend⒙w orkday/holiday⒙etc. An index m apping database is designed to show the different m apping functions for various factors. According to the basic principles ofpattern recognition⒙cluster analysis technique is used to describe the difference betw een the future day and past days. As itsapplication⒙a new short term load forecasting m ethod nam ed factor m atching m ethod is put forw ard. Case study show s thatm ore accurate forecasting results can be got by the proposed m ethod.

关 键 词：短期负荷预测 模式识别 聚类分析 电网 负荷预测 

分 类 号：TM715[电气工程—电力系统及自动化]