Aggregation Modeling of Wind Farms Based on Multi Machine Representation

摘要

Aiming at the uncertainties of wind turbines clustering results and equivalent model caused by the changes of operating scenarios in wind farms, an aggregate modeling method for wind farms based on multi-machine representation is proposed. Firstly, based on the Markov property of wind speed, the probability transfer matrix is established to characterize the seasonal and daily characteristics of wind speed. Secondly, the two-step method is used to aggregate and classify wind turbines according to wind speed and operation characteristics of front-end speed-regulating wind turbines. According to Fisher criterion, the clustering results were analyzed by probability analysis and regression test in order to merge the groups with higher correlation. Finally, the highest probability of each quarter as the final clustering results, so as to get the wind farm aggregation model. Compared with the output characteristics of the detailed model, the proposed modeling method and the aggregation model can more comprehensively characterize the operation characteristics of wind farm. Thus, the adaptability of the equivalent model is improved effectively, and the uncertainty of the clustering results of the equivalent model is solved.