摘要
分析了当前超短期负荷预测中存在的主要问题;在对大量历史负荷观测的基础上,提出并应 用聚类分析理论进行负荷变化趋势分析;通过分析,在固定分类预测算法的基础上,提出了动态分 类预测算法,该方法能够根据预测目标自动调整预测样本;大量的模拟测试表明,改进后的预测方 法能够在无需频繁维护样本的情况下,大幅提高超短期负荷预测精度,尤其是对节假日负荷预测, 效果更为明显。
This paper applies cluster analysis theory in analyzing load tendency through discussion on the main problems existing in ultra-short term load forecasting, and presents a fixed cluster load forecasting method firstly. And then an improved method, dynamic cluster load forecasting method, is proposed based on the analysis of the fixed cluster method, which could adjust input samples in terms of the target of forecasting automatically. Field test on historical system loads shows that the dynamic cluster load forecasting method proposed is superior to the methods available at present, to a great extent improving the accuracy of forecasting result without frequently maintaining forecasting samples, particularly in holiday load forecasting.
出处
《电力系统自动化》
EI
CSCD
北大核心
2005年第24期83-86,97,共5页
Automation of Electric Power Systems
关键词
超短期负荷预测
聚类分析
负荷趋势
固定分类
动态分类
实时调度
ultra short term load forecasting
cluster analysis
load tendency
fixed cluster
dynamic cluster
real-time dispatch