Long-term wind power forecasting model by multi meteorological variables based on data compensation

Long-term wind power forecasting plays an important role in the planning, prevention and control of harmful wind power events. However, short operation time of domestic wind farms result in the deficiencies of wind power data and it restricts the modeling of wind power forecasting. The staging operation of wind farms also bring bad effect to forecasting. To solve the problems, this paper proposed a long-term wind power forecasting model by multi meteorological variables based on data compensation. This model uses correlation coefficient method to select the optimal meteorological variables. Establishing the RBF neural network between optimal meteorological variables and historical power to get years of fitting power data. In this way, the power data can be compensated. Then, using the fitting power to replace the special year's wind power and forecasting the next year's wind power in steps of 3h. The verification with the actual data of Nanshan wind farm shows the proposed long-term forecasting method is effective and practicable. Those problems, deficiencies of power data and the special year, are solved.