| 引用本文: | 梁海平,田潮,王铁强,曹欣,杨晓东,刘英培.基于改进K-means算法的电网运行断面相似性匹配研究[J].电力自动化设备,2019,39(7): |
| LIANG Haiping,TIAN Chao,WANG Tieqiang,CAO Xin,YANG Xiaodong,LIU Yingpei.Running section similarity matching based on improved K-means algorithm[J].Electric Power Automation Equipment,2019,39(7): |
|
| 摘要: |
| 为简化电网工作票制定过程中复杂的方案校验工作，降低对电网调度人员工作经验的依赖，提出利用断面特征提取进行电网历史运行断面相似性匹配的方法。首先根据系统运行特点和数据存储格式，采用决策树模型提取、筛选特征变量；然后基于改进半监督K-means算法对历史运行断面进行初步相似性聚类，获取有效样本，降低数据规模；最后利用相似性匹配指标体系在聚类结果中为系统当前运行断面匹配到最有参考和利用价值的历史运行断面及其对应决策信息。仿真算例表明，所提方法可以很好地完成运行断面的相似性匹配工作。 |
| 关键词: 工作票 运行断面 半监督K-means算法 相似性匹配指标体系 聚类算法 |
| DOI：10.16081/j.issn.1006-6047.2019.07.018 |
| 分类号:TM761 |
| 基金项目:中央高校基本科研业务费专项资金资助项目(2016MS88,2017MS091)；国家电网公司科技项目(SGTYHT/17-JS-199) |
|
| Running section similarity matching based on improved K-means algorithm |
|
LIANG Haiping1, TIAN Chao1, WANG Tieqiang2, CAO Xin2, YANG Xiaodong2, LIU Yingpei1
|
|
1.School of Electrical and Electronic Engineering, North China Electric Power University, Baoding 071003, China;2.State Grid Hebei Electric Power Company, Shijiazhuang 050021, China
|
| Abstract: |
| In order to simplify the complex scheme checking work in the process of making power grid work tickets and reduce the dependence on work experience of power grid dispatchers, a running section similarity matching method of power grid by section characteristic extraction is proposed. Firstly, the decision tree model is adopted to exact and screen the characteristic variables according to the system operation characteristic and data storage format. Then, initial similarity clustering is carried out based on the improved semi-supervised K-means algorithm to obtain effective samples and reduce the data scale. Finally, a similarity matching index system is used to match the most valuable and useful historical running section and its corresponding decision information for the system current running section in the clustering results. Simulation cases show that the proposed method can well match the similarity of running sections. |
| Key words: work ticket running section semi-supervised K-means algorithm similarity matching index system clustering algorithms |
