The Dynamic Equivalence For Wind Farm Based On Clustering Algorithm

With the development of wind power generation technology, grid-connectedwind farm capacity gradually increases. It is a need to do detail research foroperation characteristics grid after a large-scale integration of wind power. Thefirst problem is that wind farm modeling, have prominent features of wind powercompared with conventional energy generation.In the practical engineering, it is usually the inflation of the grid caused bywind farm, i.e., a wind farm sum all wind turbine effects. In the modeling process,if using a detailed model of wind farm, a wind turbine and a electrical network ismodeling. The effects on the power system can be viewed as a plurality of smallcapacity of the wind turbine, a step-up transformer and a large number of connectedline models added into the power system model. It not only increase the size of themodel in the power system, but also brings many serious problems, such as thevalidity of the model and data modification.It also will increase the time of flowcalculation and especially increase domain simulation time. Therefore, in order toreduce the computation time and workload, it is necessary to take the equivalentmethod, through one or several wind turbines to describe the characterization ofwind farm.The wind generator is affected by the factors of drafting effect in the actualsetting, and running status are not identical. In order to improve the precision of themodel of the wind power at the actual operation, a dynamic equivalence modelingmethod that is suitable for doubly-fed wind generators is provised.Firstly, thedoubly-fed wind turbine steady-state and transient process are deduced from voltageequation and flux equation of doubly-fed wind generator. The state variables in thetransient process are adopted as13state variables as clustering index,and it will beinitialized through the process of steady state. The state variable matrix is formatted,and the state variables are the initial calculation. Then state variable matrix is clusteringly processed by the kmeans method of clustering algorithm, with thegroup of wind power generator equivalent to become a wind turbine wind farm. Itrealizes the dynamic equivalence. Equivalent generator parameters and equivalentfield connection parameters were calculated in accordance with the capacityweighting principle and loss principle of invariance, through calculation, comparedwith a detailed model, to test the accuracy of calculation.Using PSD/BPA platform,for faults and wind speed disturbance two circumstances are simulated. It iscontrasted with the traditional equivalent method of wind turbine and detailedmodels. The simulation results show that it is reasonable that the state variables ofsimulation process sserves as cluster-dependent index.The dynamic characteristic ofthe new model is the same to the detailed model. Characterization of actual states ofwind farm can be showed.