Short-term Wind Power Forecasting Based On Support Vector Machine

With the development of the world, the demand and consumption of energy has rapidly increased. Long-term exploitation and consumption of the non-renewable energy sources such as the traditional oil, coal and other sources means that they will be exhausted in the future. So, the search for another new energy has become an urgent demand for human development. For that reason, many countries start to develop the wind power because of its abundance and wide distribution on the earth. Due to the fluctuation and intermittence of the output power of the wind power plant, the balance of the generation, transformation, transmission and distribution of the electricity of the whole electrical power system will be destroyed, threatening the safety of the power grid. If the wind power can be forecasted accurately, it will be effectively affordable to guide the power grid to deal with the excessive wind power. As the result, it’s a great importance to do the research of the prediction technique of the wind power for a better development of the wind electricity.The main research contents of this paper are as follows:1. Using the support vector machine as the basic tool of modeling and building wind power prediction model all aim at a short-term prediction of the output power of the wind electricity system.2. There is a research for the parameter optimization of the support vector machine modeling. The selection of support vector machine modeling’s parameter affects the establishment and the accuracy of the predicted model. It is proposed that the cuckoo search algorithm instead of the traditional method will be more favorable to improve the capacity of the optimization of the support vector machine. Compared with the traditional grid parameter optimization, that method is more accurate and has high efficiency by means of the simulation results, which really improves the precision and accuracy of the forecast of the wind electric power.3. As the wind power output is influenced by some uncertain factors,and there also exists the uncertain information and noise in training sample, the use of fuzzy C mean value algorithm in fuzzy theory will besuitable to do the de-noise process, which combines the advantage of the fuzzy support vector machine theories. At the same time, applying the cuckoo algorithm for parameter optimization and using support vector machine training model accelerate the effect of the wind power forecast by means of the simulation experiment.