基于聚类趋势分析与逐步回归的电铁牵引负载负序源模型研究  

Study on Negative Sequence Source Model of Electrified Railway Traction Load Based on Clustering Tendency Analysis and Stepwise Regression

作　　者：孙谦[1] 姚建刚[1] 李欣然[1] 孔强 胡锦泓 康童 徐振超 

机构地区：[1]湖南大学电气与信息工程学院,湖南省长沙市410082 [2]湖南湖大华龙电气与信息技术有限公司,湖南省长沙市410012

出　　处：《中国电机工程学报》2012年第34期120-128,18,共9页Proceedings of the CSEE

基　　金：国家重点产业振兴和技术改造项目(国发改投资[2010]2272);高等学校博士学科点专项科研基金资助项目(20070532052)~~

摘　　要：建立适合于负序特性分析与潮流计算的电铁牵引负载负序源模型至关重要。负序源建模可视作负荷建模问题,提出一种基于负序特性分类与综合的建模方法。选取基波负序电流的实测响应空间与机车的实际运行状态为特征向量,对样本集进行聚类趋势分析。在确保样本集可聚的前提下,利用聚类有效性函数得到最优分类结果。通过对牵引负载的负序特性机理分析确定模型的结构,并采用逐步多元回归的方法得到最优模型表达式。对新增样本,将其归入与聚类中心欧氏距离最小的那个类,并对该类模型进行检验。实例分析表明,所建负序源模型是合理而有效的。To establish the electrical railway traction load negative sequence source model suitable for negative sequence characteristics analysis and power flow calculation is extremely important.Negative sequence source modeling could be regarded as a load modeling problem,and a modeling method based on the classification and synthesis of negative sequence characteristics was proposed.The clustering tendency analysis was conducted on the sample set,with the measured response space of fundamental negative sequence current and the actual running state of locomotive as eigenvectors.On the premise of clustering possibility,the clustering validity function was used,aiming at obtaining the optimal classification result of the sample set.By means of mechanism analysis for the negative sequence characteristics of traction load,the model structure was determined,while the optimal model expression was achieved applying the method of stepwise multiple regression.Besides,the newly added sample was classified into the class within the minimum Euclidean distance from the clustering center,and the related model was tested.Case study shows that the negative sequence source model built is reasonable and effective.

关 键 词：负荷建模 牵引负载 负序特性 分类与综合 聚类趋势 逐步回归 

分 类 号：TM714[电气工程—电力系统及自动化]