| With the prominent problems of energy and environment,wind power has the advantages of flexibility,cleanness and sustainability,so it occupies an increasingly high proportion in the power grid.The large-scale integration of wind power has changed the original power flow and inertia of the power system,affected the stability of the power system,and made the model identification of the power system more complex and difficult.Therefore,the establishment of wind farm equivalent model with the same characteristics as the actual wind farm and the accurate measurement and identification of the parameters of the equivalent model have become an important problem to be solved urgently in power system operation control.In this paper,the model of DFIG is established first,and the accuracy of the model is verified by low voltage ride through test data.Then,the aggregate equivalent model of DFIG-based wind farms is introduced.Finally,the identification modeling schemes of wind farms under different measurement data are studied,and a software platform for wind farm model detection is established to generate the model online.Under the condition of complete information,all the information of wind farm can be obtained.At this time,the key is to select the appropriate clustering index and clustering method to reasonably and effectively cluster the units working in different conditions in the wind farm,and equivalent each group of units into an equivalent machine,and calculate the parameters of the equivalent machine.Therefore,this paper puts forward a method for identifying the structure of the wind farm aggregation equivalent model based on the active power output of wind turbine,which realizes the measurement and identification of the number of equivalent machines.Then,the operation parameters and body parameters of the equivalent machine are identified accurately,and the body parameters of the equivalent machine are calculated by the capacity weighting method.For the equivalent parameters of collector lines,a parameter identification method based on the principle of equal current and power injected into nodes before and after equivalence is presented,which not only improves the efficiency of parameter calculation,but also ensures the accuracy.Finally,the identification scheme of equivalent model with complete information is formed,and its effectiveness is verified by an actual DFIG-based wind clusters.However,in real life,the huge real-time operation information at the unit level is difficult to obtain for the actual wind clusters.In view of the missing operation data at the unit level,this paper puts forward a solution that first establishes the mapping model between the measured data of the common connection points at the farm and the data at the unit level,then carries out the model identification.According to the historical operation data of wind farm,this paper analyzes its characteristic scenes and establishes the historical scene database of wind farm.Then,according to the scene in the database,the multi-level neural network model of wind farm is trained,and the matching model of unit level information is established.At the same time,a matching mechanism of negative feedback regulation is proposed to ensure the consistency of the total power of the wind farm and improve the accuracy of data matching.Finally,the identification scheme of equivalent model is formed under the condition of missing information,and its effectiveness is also verified by an actual DFIG-based wind clusters.After establishing the identification scheme of wind farm equivalent model under different measurement data,this paper builds the software platform of wind farm model identification through Python language,and designs four software interfaces of "scene matching","structure identification","parameter identification" and "result display",so as to realize the online generation of wind farm identification model under different measurement data. |
