Research And Application Of Short-term Wind Power Forecasting Based On Delay Optimization Strategy

The use of renewable energy is an significant way to solve the energy shortage,and wind energy has become one of the most popular renewable energy in the world with its continuous and pollution-free characteristics,but the instability of wind power generation has always been a problem.Wind energy forecasting is one of the solutions to this problem.Accurate forecasting helps to develop appropriate incentives for a well functioning electricity market.From the physical point of view,wind power generation is a process of energy conversion.When the wind speed changes,the rotation of the fan blades is affected by the mechanical inertia,and the actual output power of the fan will deviate from the ideal output power,especially when the wind speed keeps changing continuously.Therefore,this paper studies the delay effect of wind speed characteristics on the short-term wind power prediction results.The main contents are as follows:The annual data of wind turbine is integrated to fit the standard wind power curve.The wind speed is divided into upwind and downwind by the positive and negative difference of wind speed,and the influence of upwind and downwind on power is analyzed.Considering the influence of the continuous rise of wind speed on the power,the non synchronous long continuous rise wind speed segment is extracted,and the wind speed rise segment is quantified by the related definitions of amplitude,angle and power difference.The appropriate length of wind speed rise segment is selected for analysis through experiments.Combined with the physical process of wind power conversion,a delay revision function is proposed to revise the wind power prediction results.Through the analysis of RMSE and MAE error evaluation indexes,it is found that the short-term wind power prediction based on the delay revision function can effectively reduce the impact of the continuous rise of wind speed on the power prediction results.Combined with spectral clustering,the rising section of wind speed is classified and analyzed.In order to identify the number of feature vectors and automatically determine the number of clusters better.The method of feature interval averaging combined with confidence interval is proposed to determine the feature vector.The accuracy of the clustering results is analyzed by the revised results and power difference of different categories,and whether the wind speed of different categories in the rising section needs to be revised is discussed.Compared with the third chapter,it is found that the revised result is better and the probability of error increase is reduced after adding cluster analysis.On the basis of the above research,this paper analyzes the development demand of wind power forecasting system,and participates in the research and development of a short-term wind power forecasting system for a wind farm in Shanghai.It mainly involves the statistical module and prediction module,which realizes the functions of meteorological data statistics and visualization,wind field monitoring,wind speed and power prediction.