| Due to the rapid development of wind power and its increasing grid capacity in the grid,the stability of wind farm power supply quality is critical to the safety and stability of the grid.Therefore,in the actual operation,the accurate wind power prediction is important for the timely power generation scheduling and the peak of the power grid,and the flux of the wind farm.This topic has conducted in-depth research in the short-term wind power prediction of wind farm.The main research contents are as follows:Firstly,the characteristics of natural wind are analyzed,then combined with the two energy conversion processes of wind turbine in wind farm,the structure and working principle of wind turbine and wind turbine are introduced respectively.Meanwhile,the meteorological factors affecting wind power in wind farm are analyzed.Aiming at the problems of randomness and volatility in wind power prediction,a prediction method based on variational modal decomposition(VMD)and JAYA optimized minimum support vector machine(LSSVM)parameters is proposed to realize short-term wind power prediction.This prediction method analyzes the historical wind speed sequence and the influence of air pressure on the wind power,uses VMD to decompose the historical wind speed,reduces the noise of the wind speed signal,reduces the noise influence of the wind speed,and combines the decomposed wind speed component with the meteorological factors.Air pressure is used as the training input of the LSSVM prediction model.The parameters of LSSVM are optimized using the optimization characteristics of the JAYA algorithm.The JAYA algorithm has no algorithm-specific parameters,only the parameters common to the traditional algorithm,so the structure is simple,the convergence speed is fast,and the establishment is short-term Wind power prediction model.Taking the measured data of wind farms as an example,the simulation analysis is carried out.The simulation results verify the accuracy of the proposed VMD-JAYA-LSSVM forecasting method for short-term wind power forecasting.Aiming at the possible errors and omissions of the original data in wind power prediction and the randomness of wind power,a combined prediction method based on data processing(DP)and robust variational echo state network(RVESN)is proposed.This method first uses the mutual information method to determine the wind speed time series that has the greatest influence on the wind power output,and then uses the Gaussian algorithm to fit the historical wind power and wind speed time series,and correct and supplement the wrong or missing data in the original wind power.At the same time,the nuclear principal component analysis(KPCA)method is used to process the time series that affect the power output of wind power,namelywind speed,wind direction,temperature,air pressure,and air density,and extract the principal components that can reflect most of the characteristics of the original time series to eliminate data redundancy.Finally,the processed data is used as the input of the robust variational echo state network to train the prediction model,and the output weight of the model is obtained,thereby establishing a short-term prediction model of wind power.Simulation and analysis of the measured wind power of a wind farm in the eastern United States collected on the NREL platform,and compared with other forecasting methods,verifying the feasibility and higher accuracy of the proposed forecasting model.Considering the actual functional requirements of wind farm short-term power forecasting system,combined with the development status of wind power system at home and abroad,a web version short-term wind power forecasting system is developed by using software tools such as Spring Boot based on J2 EE,My Batis back-end framework,My SQL database and visual React framework based on Java Script displayed on the front end,so as to realize the visualization of wind farm short-term wind power exhibition. |
