Wind Speed Forecasting Model For Wind Farms

With the dwindlement of non-renewable energy resource, the exploitation of renewable energy resource has drawn more and more attention. Among these renewable energy resource, wind energy plays an important role in that, not only developed countries, but also many developing countries wind power has been widely used, even in some developed countries, wind power has replaced traditional power generation model to provide the basic driving force of economic development. Wind speed forecasting play an important role in the exploitation of wind power, and assessment of wind resources, grid-connected and planning of wind farm, power system and so on. So wind speed forecasting has an important significance to the development of wind power.In this paper, it introduces the current status of wind speed forecasting at first, and summarizes the main methods for forecasting. In neural network, the ability of approximation of nonlinear function is used to design three kinds of single neural network forecasting model, and their basic principles and realization of each model are analyzed. The current focus on wind speed forecasting model is single model, so in this paper, based on the research on single forecasting models for wind speed, combination forecasting models are proposed to predict wind speed. Unlike one-step forecasting of single model, combination forecasting model is achieved through multi-step, which combines the information of other models, and can effectively use the various relations among data, to reach the demand of improving forecasting accuracy.There is some difference between combination forecasting model in this paper and traditional linear combined forecasting model, the combination forecasting model is a nonlinear forecasting model. The forecasting wind speed of all single models as its inputs, and the actual wind speed as its outputs, is constructed based on neural network, and trains the network to meet the error demand, then the model can be used to complete the final wind speed forecasting. In order to compare the advantages and disadvantages of the combination model, linear combined forecasting model is established to predict wind speed. Three error indicators are used to evaluate the forecasting accuracy of all single forecasting models and combination models. This forecasting way is more complex than one-step way, but to some extent it can overcome the problem of incomplete learning to data. The results of simulation and error analysis show that the combination forecasting model based on neural network can effectively improve the prediction accuracy of wind speed forecasting, and proved the feasibility of the model, and provides some ideas for further promotion of forecasting accuracy for wind speed.