基于广域信息的同调机群聚类识别方法  

A Wide Area Information Based Clustering Recognition Method of Coherent Generators

作　　者：张亚洲[1] 张艳霞[1] 蒙高鹏 赵冉[1] 高瑾[1] 

机构地区：[1]智能电网教育部重点实验室(天津大学),天津市南开区300072

出　　处：《电网技术》2015年第10期2889-2893,共5页Power System Technology

摘　　要：针对目前基于广域测量系统(wide area measurement system,WAMS)的发电机同调分群聚类方法仅仅以功角曲线之间的距离作为聚类指标,忽略了轨迹的局部特征及变化规律,导致轨迹分析不全面的问题,提出了一种基于轨迹结构差异度的同调机群聚类算法。首次将通信领域的轨迹特征提取方法应用到电力系统中,较全面、综合地考虑了发电机功角曲线之间的距离、变化方向、波动程度和变化速度的差异,利用变异系数法进行特征权重计算,以曲线之间的结构差异度为聚类指标实现了多机系统的同调分群。通过EPRI-36节点系统仿真算例验证了方法的有效性,结果表明:算法可以根据电力系统的具体运行情况确定合适的权重值对发电机之间的特征差异进行评价。与已有的基于功角轨迹的分群方法相比,轨迹分析更全面,分群结果更具有实际意义。Existing generator coherency cluster grouping methods based on WAMS only regard distances between power angle curves as clustering index, ignoring local characteristics and varying trends of power angle tracks, therefore lead to problems of incomprehensive trajectory analysis. This paper presents a coherent clustering algorithm based on diversity of track structures. The method of extracting trajectory characteristics in communication field is applied to power system. This method takes comprehensive considerations of all kinds of differences, including distance between generator power angle curves, changing direction, degree of volatility and changing speed of power angle tracks, uses variation coefficient method to calculate feature weights, takes structure difference between curves as clustering index, and precisely realizes coherency grouping of multi-machine systems. Effectiveness of the proposed method is verified with EPRI-36 node system simulation example. Analytical results show that appropriate weight values to evaluate difference between characteristics of generators can be determined according to specific operating situation of power system. Comparing with existing clustering methods based on power angle trajectory, this algorithm is more comprehensive and its clustering result has more practical significance.

关 键 词：广域测量系统 同调识别 轨迹聚类 结构差异度 

分 类 号：TM74[电气工程—电力系统及自动化]