An intelligent wind farm model for three-phase unbalanced power flow studies

With the rapid growth of wind power penetration in power systems, researchers focus on methods to accurately model wind generators in power flow studies. There are several accurate wind generator models which capture the voltage dependence of power output per each phase of wind generators. These models have been built using individual models of all the constituent components of wind generators. Furthermore, these models comprise complex nonlinear equations and hence inevitably slow down power flow studies. When wind farms are modeled with this approach, they become very complex and cumbersome to be integrated into power flow studies. On the other hand if the power output of a wind farms is simplistically assumed as fixed injection value neglecting the voltage dependence of power output per phase, the resultant power flow solution will not be accurate due to over simplification. In this paper a new wind farm model is built using Artificial Neural Networks (ANN). The procedure of building ANN models is explained using a small wind farm with five wind generators. The ANN wind farm models estimate power output per phase using three-phase voltages and wind speeds. A power flow study with this ANN model, a simple fixed power model and a detailed nonlinear model is reported in this paper with sufficient comparisons. The proposed ANN model is 80 times faster than a complete nonlinear wind farm model and as accurate as the nonlinear wind farm model.