暂态稳定分析多机等值时,受扰发电机的分群是需要解决的首要问题。以福建电网为研究对象,建立了同调机群的数学模型,针对机组分群特点对传统聚类评价标准之一的曼哈顿距离进行了优化,提出了一种全新的基于发电机功角轨迹的相似度函数作为聚类规则;对于大电网机组的分群,提出了一种基于主导机组的研究思路,并运用BPA软件对出现在福建电网不同区域的大扰动和小扰动分别进行了仿真分析,利用层次聚类法对电网机组进行了聚类分群,有效验证了该聚类分群方法的可靠性,并对福建电网机组分群规律进行了总结。 Transient stability analysis of multi-machine equivalent, the disturbance of the generator grouping is the primary problem to be solved. Taking Fujian Power Grid as the research object, the mathematical model of the cohomology cluster is established, and the Manhattan distance which is one of the traditional clustering evaluation criteria is optimized according to the characteristics of the cluster. A new similarity function based on the generator power angle trajectory is proposed As a clustering rule. For the grouping of large power units, a research approach based on leading units is proposed. BPA software is used to simulate the large disturbances and small disturbances that occur in different regions of Fujian power grid. By using hierarchical clustering The method of clustering and grouping of grid units validates the reliability of the cluster clustering method and summarizes the rules of grid grouping in Fujian grid.