Short-term Prediction Research Of Wind Power Based On Fuzzy Clustering Algorithm And GA-BP

With the strengthening of people’s environmental awareness, wind energy is paid more attention and got more utilization as a new energy which is of widely used and maturely technology. Striving to develop wind energy becomes a major objective. In order to economize wind power reasonable, the difficulties and problems to solve should be taken into consideration. It is widely known that wind power is of randomness and volatility. These would bring about the violent fluctuations of wind power connected into power grid, which results in a series of problems like frequency, power angle, voltage stability and problems such as harmonic waves, flickering, transmission losses and system reliability. These problems can not only influence power system, but also destroy the economic, secure, stable and reliable operating state of electrical power system.To avoid the wind power grid-connected problems, the power forecast in wind farm should be much precise so as to the operating crew can formulate schedule, arrange sets output and system reserve to reduce the impact brought by wind power grid-connection.However, it is usually hard to exactly predict because there are many factors to affect the wind speed. So only continually improvement on algorithm study, modeling and useful information collection can increase the prediction accuracy.The historical data of one wind farm is used as original data and intuitional contrastive analysis of these models’ prediction accuracy are finished after their practice and forecast. In this paper, single BP neural network, single radial basis function neural network and the BP neural network after genetic algorithm optimization are used to predict wind speed.After the analysis of the three methods, the wind speed and wind power short-term prediction research based on fuzzy clustering algorithm and GA-BP (the BP neural network after genetic algorithm optimization) is put forward and which is taken meteorological factor into consideration. First, historical samples’ classification is made by fuzzy clustering technology for there is no need to get accurate results in wind speed prediction. Second, historical data after fuzzy clustering algorithm processing is served as input of GA-BP neural network model and the prediction results are obtained, thus wind power is acquired through the power curve of wind power generator. The prediction results’ effectiveness is estimated according to the actual measurement data of the wind farm, and then the results are compared with which obtained from single BP neural network model and BP neural network model after genetic algorithm optimization.