中国电力 ›› 2016, Vol. 49 ›› Issue (9): 137-141.DOI: 10.11930/j.issn.1004-9649.2016.09.137.05

• 技术经济 • 上一篇    下一篇

基于数据均值化及LSSVM算法的峰谷电价需求响应模型

李娜1, 张文月2, 王玉玮3, 符景帅3, 王炜劼2, 王麟2   

  1. 1. 国网天津市电力公司经济技术研究院,天津 300171;
    2. 国网天津市电力公司,天津 300010;
    3.华北电力大学,北京 102206
  • 收稿日期:2016-05-05 出版日期:2016-09-10 发布日期:2016-09-28
  • 作者简介:李娜(1985—),女,天津人,博士,经济师,从事电力市场理论及应用技术研究。E-mail: linajamie@163.com

Demand Response Model of TOU Electricity Price Based on Data Mean and LSSVM Algorithm

LI Na1, ZHANG Wenyue2, WANG Yuwei3, FU Jingshuai3, WANG Weijie2, WANG Lin2   

  1. 1. State Grid Tianjin Economic Research Institute, Tianjin 300171, China;
    2. State Grid Tianjin Electric Power Company,Tianjin 300010, China;
    3. North China Electric Power University, Beijing 102206, China
  • Received:2016-05-05 Online:2016-09-10 Published:2016-09-28

摘要: 为提高系统的鲁棒性,客观挖掘特定用户对峰谷分时电价的需求响应规律,采用历史数据均值化的预处理方法,在一定程度上去除历史数据中其他非价格因素随机波动对用户需求响应行为的影响;运用构造等效峰谷分时电价思路,对训练样本集中的数据容量进行扩展;运用最小二乘支持向量机回归技术在所构造的训练样本集上建立分时电价需求响应预测模型;以地区T某行业为例进行算例仿真,在验证模型合理性的基础上,得到该行业日峰荷、平荷、谷荷对分时电价变动的响应规律曲线,该曲线符合用户峰谷分时电价需求响应曲线的实际情况,从而进一步验证了模型的可行性。

关键词: 峰谷分时电价, 需求响应, LSSVM, 数据挖掘

Abstract: In order to improve system robustness and meet demand of customers’ behavior rule under time-of-use (TOU) price, a simulation model is proposed. Firstly, with historical data equalization, the impact of non-price fluctuations in historical data on user’s demand response is removed. Secondly, training sample set data capacity is extended by establishment of equivalent TOU. Thirdly, a forecasting model is constructed by using LSSVM regression technique. Lastly, by using historical data of an industry in T area, the demand response curves of Peak, Flat and Valley Load under TOU are obtained to verify validity of proposed model.

Key words: time-of-use price, demand response, LSSVM, data mining

中图分类号: