In recent years, wind power generation forecasting technology has received much attention. However, because many wind farms are located in complex terrains and environments, it is difficult to eliminate the impact of microclimate on wind power forecasting and improve forecast accuracy. Long-term forecasting of wind power has become an urgent problem that needs to be solved.
The impact of microclimate in the wind power forecasting process mainly comes from two aspects: the gap between the micro-scale wind field modeling and the actual environment and the inability of digital weather simulation to refine the local microclimate of wind energy. wind field.
In order to eliminate the impact of microclimate on forecast accuracy, technical implementation is carried out in two aspects: correction of the CFD micromodel on the basisactual measured data from the wind tower; statistical correction based on wind field; operating data. The applied statistical correction adopts two different methods, and their effects are compared: the linear statistical correction method combined with the downscaling model design and the neural network method to correct the operating data; This study combined the analysis of wind farms in a complex mountainous forest area in Heilongjiang to carry out a comparison of different correction methods, and further clarified the technical route of combining CFD technology and using neural network methods to correction in order to eliminate the impact of the microclimate on electricity. forecasting production power. Efficiency and superiority improve the accuracy of actual wind farm forecasts.