Probabilistic computation of a wind farm generation based on wind turbine dynamic modeling

This paper addresses the problem of predicting a wind farm's power generation when no or few statistical data is available. The study is based on a time-series wind speed model and on a simple dynamic model of a DFIG wind turbine including cut-off and cut-in behaviours. The wind turbine is modeled as a stochastic hybrid system with three operation modes. Numerical results, obtained using Monte-Carlo simulations, provide the annual distribution of a wind farm's active power generation. For different numbers of wind turbines, we compare the numerical results obtained using the dynamic model with those obtained considering the wind turbine's steady-state power curve. Simulations show that the wind turbine's dynamics do not need to be considered for analyzing the annual distribution of a wind farm generation.