基于自组织映射神经网络K-means聚类算法的风电场多机等值建模

Multi-machine Equivalent Modeling of Wind Farms Using SOM-based K-means Clustering

为研究高比例风电接入对电网安全稳定性的影响,提出了基于SOM(自组织映射神经网络)K-means聚类的风电场多机等值建模方法。首先选取风电场运行时的有功功率、无功功率、机端电压、输出电流、平均风速5种状态变量作为聚类算法的输入变量矩阵,通过基于SOM K-means聚类算法对变量矩阵进行处理,得到风电机组等值群数。然后用1台机组并联理想受控电流源的方法表征整个同群机组,得到风电机组的多机等值模型并进行仿真计算。最后通过与单机模型及详细模型在风速扰动和短路下的仿真曲线对比验证所提出的多机等值方法的有效性。

In order to study the impact of high proportion wind power integration on grid security and stability,a multi-machine equivalent modeling method for wind farm using SOM-based(self-organizing map)K-means clustering is proposed.In this paper,active power,reactive power,terminal voltage,output current and average wind speed are selected as input variable matrix;then the variable matrix is handled through SOM-based K-means clustering to conclude equivalent group number of wind turbines.The same group of wind generators are represented by an equivalent wind turbine through a wind generator paralleled with an ideal current source to conclude a multi-machine equivalent model and simulation.Through comparison of simulation curves of single-machine model and detailed model under wind speed disturbance and short-circuit fault,the multi-machine equivalent method is verified.