Study On Combination Forecasting For Shotr-Term Wind Power Based On Artificial Intelligence

With the rapid development of society and economy, the traditional fossil fuels can not meet the demand for energy, various countries have focused on the development of renewable energy in which wind power has been highly concerned. In recent years, wind power has developed rapidly in China, according to the statistics of Energy Bureau, the wind power generation accounts for 3.3% of total generated energy in 2015. Along with the continuous increase of the proportion of wind power, wind power integration has a serious impact on safe and stable operation of Power Grid and power dispatch, accurate short-term wind power prediction can effectively solve the problems. Therefore, the study of short-term wind power forecasting has great significance.(1) The basic theory of short-term wind power forecasting is introduced based on the basic method of short-term wind power forecasting, error analysis and process of forecasting mechanism. Then, an example is analyzed by respectively using forecasting method based on similar day method and LS-SVM(Least Squares-Support Vector Machine) method, the forecasting results show that the LS-SVM method is better than the similar day method.(2) To overcome the defects of traditional clustering algorithm, a comprehensive clustering algorithm is proposed based on improved fuzzy C-means clustering algorithm, and it is used to achieve the center of the RBF neural network, then, a short-term wind power forecasting method based on RBF neural network is given. Finally, an example is analyzed by used the RBF neural network, and the results show that the method can effectively improve forecasting precision.(3) Aiming at the limitations of single forecasting method, two basic methods of combination forecasting are explained, and an optimal combination forecasting method for short-term wind power forecasting is given by using combination forecasting method based on IOWGA operator for degree of logarithm grey incidence. Finally, the RBF neural network method, similar day method and the LS-SVM method are considered to single forecasting method, and an example is analyzed by using the optimal combination forecasting method and the two basic methods, the results show that the optimal combination forecasting method can effectively integrate the information of all single methods to improve forecasting precision.