A New Method of Wind Farm Clustering Based on Improved Fuzzy C-Means Clustering and MPSO

Abstract

Affected by many factors such as irregular cluster distribution, complex terrain and wake effect, the operation conditions of each unit in large-scale doubly fed wind farm are different. In order to improve the accuracy of equivalent output model of wind farm, an improved equivalent modeling method of doubly fed wind farm is proposed. Firstly, the characteristic state variable matrix which can represent the operation state of each unit is selected as the clustering index, and the improved fuzzy c-means clustering algorithm is used to divide the cluster, then the parameters of the peer check-in model are identified based on the global optimal position variation particle swarm optimization algorithm, and finally the same group of units is equivalent to a fan. DIgSILENT simulation software is used for modeling, two dynamic conditions of wind speed step and three-phase short circuit fault. The simulation results show that the dynamic characteristics of the equivalent model are basically the same as that of the detailed model. Compared with the traditional single machine equivalent model, the accuracy of this method is higher.