摘要
提出了用于中长期负荷预测的多重可拓聚类模型和算法。首先在构建可拓经典域和节域的基础上,结合数据聚类的相关理论,针对待预测样本的数据进行样本类别的划分;然后建立相应的关联函数并通过计算得到每个类别对应的关联度,进而决定待预测样本的所属类别;再根据所属的类别重新划分二重类别,得到更加精确的预测结果。预测实例表明,多重可拓聚类预测模型具有较高的预测精度。
Aiming at thesituation that the load forecasting of power system mais have certain errors, double extension gather-type model and calculating way are put forward. Firstly, on the bass of constructing the classic area and stanza area, combining data-gather theory, samples are divided to several types according to the data of samples to be predicted. Then homologous connection-function is built up and each connection degree of category is got by calculation, and the category which belongs to samples to be estimated is decided. Then the double category is divided again according to the corresponding category, and higher accurate results are got. The results of load forecasting demonstrate that the double extension gather-type model hase higher estimation accuracy.
出处
《电力学报》
2006年第4期439-443,共5页
Journal of Electric Power
关键词
多重可拓聚类
负荷预测
物元模型
double extension gather-type
load forecasting
matter-element model
作者简介
孙奇（1982-），男，江苏常州人，硕士研究生，电力系统负荷预测、电力系统继电保护和配电网自动化；
杨伟（1965-），男，江苏徐州人，副教授，电力系统运行、控制，电力市场。
