Wind,Solar And Water Probability Density Prediction Method Based On Fuzzy Information Granulation And Support Vector Quantile Regression

At present,Chinese power generation mode has always been dominated by traditional thermal power generation.Since the 21^st century,the total amount of electricity used in Chinese life,industry and national defense has been gradually increasing,which has brought enormous pressure and challenges to thermal power generation.At the same time,the pressure on environmental protection is growing,which forces us to seek cleaner and more efficient energy sources as soon as possible to solve the many drawbacks of traditional power generation.Wind power generation is one of the new energy sources that are widely studied and used at present,but due to the uncertainties and intermittent characteristics of wind energy,wind power generation is difficult to stably integrate into the power grid.In recent years,solar photovoltaic power generation has developed rapidly.Various countries and regions have invested huge manpower,material resources and financial resources.Through large-scale advantages and automation and intelligent equipment of related industries,photovoltaic power generation technology has developed rapidly and its market.The popularity is also getting higher and higher.Hydropower has only a small impact on the environment,low power generation cost and fast start-up speed,easy adjustment and control,and has been widely used.In order to effectively measure the uncertainty and stability of wind power,photovoltaic and runoff,and further improve the accuracy of prediction results of new energy generation power,this paper will support vector machine?SVM?and quantile according to the time series characteristics of wind,light and water.Combining the regression?QR?method with fuzzy information granulation?FIG?,a model based on fuzzy information granulation and support vector quantile regression?FIG-SVQR?is constructed.Combined with the nuclear density estimation,the probability density prediction of wind power,photoelectricity and runoff is obtained,and the more accurate wind power,photoelectric and runoff fluctuation curves and probability density curves are obtained.The FIG-SVQR prediction model is compared with the traditional neural network model?Elman,BP,RBF?and SVQR model,and analyzed with corresponding evaluation indicators.In order to verify the efficiency and stability of the proposed method,the mode and median of the probability density curve were selected as the point prediction results,and the prediction interval evaluation criteria were used to evaluate the prediction interval.In this paper,three different data sets of wind,light and water are used to analyze the case.The results show that the proposed fuzzy density granulation and support vector quantile regression probability density prediction method can obtain complete wind,light and water probability density curves.And the prediction interval better solves the volatility and uncertainty of new energy,and provides relevant technical support and theoretical basis for the stable operation of the power system.