Research On Multi-scale Dynamics Aggregated Modeling Of Wind Farms

As the construction of wind farm is upsized and in large-scale, and the proportion of wind power in electricity structure is higher, wind farm is no longer a single power technology. The connection between wind farm and electrical power system is closer and closer and mutual influence is more and more complicated. In order to accurately analyze the dynamic characteristics of wind farm and the power system which wind farm has been connected to, the model of wind farm should be deeply studied. Based on the time-scale of variable speed wind turbine generator and space-scale of wind farm, this paper sets up multi-scale dynamics aggregated modeling of wind farm.First of all, this paper proposes multi time-scale modeling for variable speed wind power plant which is dividing wind power plant modeling into minute level time-scale and second level time-scale based on its dynamic behavior. By combing mathematical models and operation control feature of variable speed wind turbine generator, the control strategies under several time dimension of the wind turbine are then presented, their performance being assessed and analyzed by means of simulations in DIgSILENT/Power Factory simulation platform. The simulation results show that the established model is correct and possesses good dynamic characteristics.Then, wind speed distribution model is built by taking wake effect into account and the formation of wind speed distribution matrix of different time is achieved based on MATLAB programming to provide parameter matrix for wind speed input module, and to provide data sample for generator space-scale division.At last, variable space-scale dynamic aggregated modeling of wind farm is set up. By analyzing the operation feature of wind power generator, this paper proposes space-scale dividing indicators, which formed by wind speed, active power, reactive power reference value, unit type and output voltage. Based on these indicators, the wind turbines are classified by using improved k-means clustering algorithm. Then, the multi space-scale clustering model of wind farm is built by combining with calculation results above. The simulation result indicates that multi-scale dynamic aggregated wind farm model set up in this paper is much closer to the dynamic response of actual wind farm,and could reflects the output features of the wind farm very well.