In order to effectively aggregate power grid demand side resources and make rational and efficient use of power grid resources, a dynamic clustering algorithm of power grid demand side resources based on dis-tributed k-means is proposed. The collected power grid demand side resource data are clustered by the distrib-uted k-means algorithm based on the confidence radius. In the fuzzy mean evolutionary neural network, the power grid demand side resource data obtained by clustering is used as the input vector to output the power grid demand side resource scenario. According to the scenario existence probability, the daily average peak valley difference of power grid side resources is the smallest taking the maximum consumption degree and the minimum daily average load fluctuation rate as the objective function and the complementarity of power grid demand side resource curve fluctuation rate and load as the constraint conditions, a multi scene clustering model of power grid demand side resources is constructed. After the model is solved by dynamically changing the inertia factor (DCW) particle swarm optimization algorithm, the multi scene clustering of power grid de-mand side resources is realized. The experiment results show that this method can realize the dynamic cluste-ring of power grid demand side resources. When this method is used to cluster power grid demand side re-sources in different scenarios, the daily load rate is low, the clustering effect is good, and can meet the needs of dynamic clustering of power demand side resources.
关键词
电网需求/侧资源/动态聚类/分布式/k-means算法/聚类模型
Key words
power grid demand/side resources/dynamic clustering/distributed/k-means algorithm/aggregation model
维普
万方数据