• CSCD核心库收录期刊
  • 中文核心期刊
  • 中国科技核心期刊

电力建设 ›› 2016, Vol. 37 ›› Issue (6): 96-102.doi: 10.3969/j.issn.1000-7229.2016.06.014

• 输配电技术 • 上一篇    下一篇

结合负荷形态指标的电力负荷曲线两步聚类算法

彭勃1,张逸2,熊军3,董树锋1,李永杰1   

  1. 1.浙江大学电气工程学院,杭州市 310027;2.国网福建省电力有限公司电力科学研究院,福州市 350007;3.国网厦门供电公司,福建省厦门市 361000
  • 出版日期:2016-06-01
  • 作者简介:彭勃(1991),男,硕士研究生,主要研究方向为电力需求侧管理,电力系统数据挖掘以及负荷预测; 张逸(1984),男,博士,高级工程师,主要研究方向为电能质量,分布式能源以及主动配电网; 熊军(1979),男,博士,高级工程师,主要研究方向为配电自动化以及智能电网; 董树锋(1982),男,博士,副教授,主要研究方向为电力系统状态估计及优化运行; 李永杰(1989),男,博士研究生,主要研究方向为电力系统小干扰稳定分析。
  • 基金资助:

    国家高技术研究发展计划项目(863计划)(2014AA051901)

A Two-Step Clustering Algorithm Combined with Load Shape Index for Power Load Curve

PENG Bo1, ZHANG Yi2,XIONG Jun3,DONG Shufeng1,LI Yongjie1   

  1. 1. College of Electrical Engineering, Zhejiang University, Hangzhou 310027, China;2. Electric Power Research Institute, State Grid Fujian Electric Power Company, Fuzhou 350007, China;3.State Grid Xiamen Electric Power Supply Company,Xiamen 361000,Fujian Province, China
  • Online:2016-06-01
  • Supported by:

    Project supported by the National High Technology Research and Development of China(863 Program)(2014AA051901)

摘要:

为改善基于欧式距离的全维度负荷曲线聚类算法在负荷形态相似度上的不足,提出了结合负荷形态特征指标的电力系统负荷曲线两步聚类算法。算法第一步采用基于欧式距离的负荷曲线聚类方法获得初步聚类结果,并通过负荷聚类评价指标选取一次聚类算法和聚类数目;第二步基于负荷形态特征指标采用监督学习算法对负荷进行重新分类。之后比较了不同算法的分类效果,最后给出了聚类结果的应用建议。算例结果表明,所提出的两步聚类算法可以改善传统的负荷曲线聚类方法在形态相似度上的不足,在二次分类方法中,支持向量机(support vector machine,SVM)算法表现较好,所提出的方法具有实际应用意义。

关键词: 负荷聚类, 电力数据挖掘, 负荷形态, 监督学习算法

Abstract:

To make up for the drawback that clustering method based on Euclidean Distance considering all dimensions of load curves is weak in load shape similarity, this paper proposes a two-step clustering algorithm combined with load shape characteristic index for power load curve. First, this algorithm obtains the preliminary clustering result by using clustering method based on the Euclidean Distance of load curves and selects the clustering method and number through cluster evaluation index. Second, this algorithm uses supervised learning algorithm to reclassify load based on load shape characteristic index. Then, different clustering algorithms are compared. At last, we propose the suggestions for the application of the clustering results. The example results show that the proposed two-step clustering algorithm can make up for the weakness of traditional clustering algorithm in load shape similarity. Support vector machine (SVM) algorithm has better performance in the second-step classification. The proposed algorithm has practical significance.

Key words:  load clustering, data mining for power system, load shape, supervised learning algorithm

中图分类号: