Study Of Short-term Wind Power Prediction

With the rapid development of the world economy and society, the traditional energy structure can not meet the people’s demand for energy, people’s attention has shifted from traditional fossil fuels to renewable and clean energy, wind energy is the highest degree of development in the all renewable energy sources. With the continuous expansion of wind power capacity, the wind power industry is facing the problem of Grid connected wind power, wind machine maintenance, improve the utilization rate of wind energy and so on,more and more attention has been paid for rational, safe and effective use of wind power, wind power prediction can effectively solve this problem. This paper summarized the development situation of wind power at home and abroad, the research situation, basic principle and forecast method of wind power prediction and the wind power prediction system has been developed to run at home and abroad.The generation of short time wind power prediction error is mainly due to the inherent random factors and external random factors. Inherent random factors is the forecasting system itself defects or imperfections, external random factors is the system input data imperfections or data inherent error. In this paper, systematic and comprehensive analyze of the effects of random factors on short-term wind power prediction, to solve the problem that the short-term wind power prediction error brought by random factors, improved and design of new prediction systems.For the problem of inherent random factors, this paper established BP neural network short time wind power prediction model, discusses determine the parameters of BP network model and the selection of hidden layers in the modeling process, Finally get the optimal BP neural network prediction model. The simulation results show that the accuracy and stability of BP network prediction is poor. In view of the BP neural network is easy to fall into local minima, the problem of poor stability, establish GASABP short time wind power forecast model that improved BP neural network model by the genetic algorithm and simulated annealing algorithm. The simulation results show that the prediction accuracy and stability are improved obviously, effectively solve the problem of local minimum.For the limitations of traditional machine study theory, establish based on statistical learning theory support vector machine wind power forecast model, use the cross grid search method to select the model parameters. The simulation results show that the prediction accuracy of support vector machine model is significantly higher than that of GASABP model. Through the comparison of the three models, through continuous optimization of forecasting model, can effectively reduce the affecting of inherent random factors on forecasting precision of wind power For the problem of external random factors, According to the above-mentioned prediction model in the influencing factors of wind power determined is no unified guiding principle, the problem that error caused by through the human according to experience to determine, determine influencing factors of wind power is not perfect. In addition, the weather forecast system of domestic wind farm is not perfect and precise, data collection will contain errors. This paper establish chaotic time series support vector machine short time wind power forecast model, by means of genetic algorithm combinatorial optimize the prediction model parameters.The simulation results show that chaos support vector machine prediction model shows good prediction performance.Compared with the support vector machine model, Chaotic time series can include all statistical rules carried by all the influencing factors. Chaotic time series support vector machine wind power forecast model can effectively reduce the affecting of external random factors on forecasting precision of wind power...