A Novel Method for Wind Farm Equivalence Based on Multi-Objective O Ptimization

Abstract

With the increase of wind power in the power grid, the impact of wind farms (WFs) cannot be ignored. For power system analysis, detail modeling a large-scale wind farms is almost impossible and unnecessary. Establish an equivalent model of wind farms has become a vital research work. The main disturbance in a WF is wind speed fluctuation and fault condition. Considering the time scale of the two kinds of disturbance are quite different, the equivalent model obtained by one disturbance may not work effectively for another disturbance. In order to solve this problem, a novel method based on multi-objective optimization has been presented in this paper. Firstly, Self-Organizing Maps (SOM) neural network used to group the wind turbines (WTs). Secondly, based on adopt trajectory sensitivity method, find out the key parameters of WT, so that the equivalent model dimensions are reduced. Finally, the key parameters of WTs are identified using Non Dominated Sorting Genetic Algorithm-II (NSGA-II). Based on the equivalent modeling analysis and experiment of a wind farm with 66 wind turbines in the northwest part of China, it is verified that the proposed equivalent model can work effectively under both wind speed fluctuation and fault condition.