Application Of Set Pair Analysis Theory In Real-time Wind Power Forecast

The development of electricity has greatly promoted the progress of human society.However,finding alternative renewable energy has become an urgent need with the reduction of fossil fuels which can generate electricity,coupled with the environmental problems caused by burning fossil fuels.Driven by this urgent demand,some progress has been made in the development of renewable energy sources such as wind power generation,solar power generation,tidal power generation and bio-energy.Among them,wind power has the best exploitation technology.In recent years,the development of wind power is greatly rapid.However,the inherent randomness and fluctuations of wind power have caused a lot of problems for the grid with the increasing scale of wind power,these features even bring severe problems to the safety and stable operation of power system.Predicting wind power accurately is a efficacious method to reduce the impact of wind power fluctuations.In order to uncover the nature of wind power fluctuations,an index named the sampling loss rate is established to measure wind power fluctuation,and the validity of this index is verified by the prediction results and the wind power smoothing effect of the wind farm.Rationality of wind power prediction results can be evaluate by sampling loss rate.Secondly,a real-time wind power prediction model is established.The key to establish an accurate wind power prediction model is to grasp the wind power trends accurately.Taking into account the influence of noise on the trend of wind power,EEMD is used to eliminate the noise of the original wind power firstly in this paper,after that,an improved rank set analysis method is used to predict the denoised wind power,and then the real-time wind power prediction model based on EEMD denoising and improved rank set pair analysis is established.Finally,the simulation experiment is carried out.The established model shows excellent prediction performance compared with some other wind power prediction models.And the universality of the model is verified by the wind power prediction results of three wind farms with different installed capacity.After that,the relationship between the wind power fluctuations and the wind power prediction error is analyzed.The reasonableness of the existing wind power prediction standard is analyzed by examples,and we find out that it is unfair to use the unified criteria to constrain different wind farms without taking account for the fluctuation of different wind farms.And we proposed that the sample loss rate can be used to develop a more reasonable wind power prediction standards.Finally,a real-time forecasting platform for wind power,real-time wind power forecasting system,is established.The system has the functions of data query,real-time prediction,error analysis,fluctuation analysis and so on,which makes the entire work of this paper more practical.