Sutudy On Ultra-short-term Combined Forecasting Considering Wind Farm Power Ramp Events

Due to the change of energy structure in China,renewable energy,as a clean and environment-friendly natural resource,has become the focus of current development.At present,China is vigorously developing renewable energy and plans to achieve "high proportion of renewable energy development" by 2050,with renewable energy accounting for more than 65%of energy consumption and more than 85% of total power generation by 2050.Therefore,as large-scale wind power generation is connected to the power system,accurate wind farm power prediction is of great significance for the safe and stable operation of the entire power system.This paper selects the historical power data of the actual operation of large wind farms to analyze and study the ultra-short-term power prediction of wind farms.First of all,because the traditional prediction in the ultra-short-term period usually only adopts the method of time series extrapolation,the error will rise rapidly with the increase of time span,especially in the scenario of large time span and wind speed mutation,ignoring the wind speed information will lead to the corresponding error.Therefore,this paper introduces numerical weather forecasting into ultra-short-term wind power forecasting and proposes a ultra-short-term wind power forecasting method considering source correlation.NWP information and time window were used to approximate the time point with low precision of rolling ultra-short-term wind power prediction,and a combination method combining neural network and persistence method was proposed to predict future wind power output.This method can make full use of NWP information and choose different prediction methods under different conditions,which has certain practicability.Secondly,when wind speed data changes rapidly,wind power ramp events caused by wind power are an important cause of wind power prediction errors.Therefore,it is increasingly urgent to consider the prediction of wind power ramp events.In order to improve the adaptability of the forecasting model under the fluctuating wind conditions,an ultra-short-term combined forecasting model considering the power ramp of wind farm is proposed based on the theory of limit learning mechanism.This method makes full use of the effective weather factors in numerical weather forecast,so that wind power can be tracked and predicted well in the event of fluctuating and irregular wind power ramp events.The results show that the method can accurately identify the power ramp events of the wind farm and effectively improve the accuracy of the ultra-short-term prediction of wind power,it has certain theoretical significance and practical value.Finally,the sources of errors are analyzed,and the prediction model is modified by using similar wind power ramp events combined with the preliminary prediction of ultra-short-term combination considering wind farm power ramp.The tuple vector time warp algorithm was used to find the similar wind power ramp events and the wind power ramp events in the prediction section,and the matching set with high similarity was obtained,so as to obtain further modified prediction results.The numerical example shows that the modified method can reduce the adverse effects of wind farm power ramp events on the prediction results,and improve the prediction accuracy effectively,which verifies the effectiveness of the proposed method.