| With the expanding scale of wind farms, the impact of stable operation of electric power system, which be caused by volatility of wind power output, is becoming more and more obvious. Therefore, it is necessary to analyze and assess the grid-connected wind farms, to grasp the safety risks and economic burden of power system operation result from the grid-connected wind farms, and build accurate model of wind farm is the base of studying the issues generated by grid-connected wind farms.In this paper, based on statistical analysis and cluster analysis of the measured operational data of wind farm, built the steady-state output model and the dynamic equivalent model of wind farm, which can reflect the actual operation situation of wind farm.Through the mathematical statistical analysis of measured wind speed data and power data the of WTGs of actual wind farm, established the wind turbine’s wind speed-power characteristic curve. Considering differences of wind velocity distribution of the large-scale wind farm may be the irregular terrain or units arranged irregularly, using a clustering algorithm to cluster the measured wind speed data of all WTGs of wind farm, built an improved wind speed equivalent model and wind farm steady-state equivalent model, which reflect the differences of wind speed distribution of the large-scale wind farm. The actual operational data of wind farm as the standard, compare the errors of traditional wind farm equivalent model and the wind farm equivalent model based on actual operational data, to verify the effectiveness of the proposed wind farm equivalent model.In order to aggregate wind farms with complex terrain or irregular layout, a new wind turbines clustering method is proposed, which through clustering the measured operational data of WTGs employing spectral clustering algorithm to cluster the TWGs. Then, the wind farm’s dynamic modeling method based on spectral clustering algorithm is given. An actual wind farm as an example, group the wind turbines using spectral clustering algorithm and built wind farm dynamic equivalent model. Compare with the simulation results of the traditional wind farm equivalent model and the proposed wind farm equivalent model, analyzing the influences of wake effect and time delay effect on wind farm’s power output, as well as differences of dynamic response of two models under wind disturbance and short-circuit fault. |
