SHORT-TERM ZONING POWER PREDICTION OF OFFSHORE WIND POWER CONSIDERING CLIMBING FEATURE QUANTITIES
Considering complex marine conditions,this paper proposes a zonal hybrid prediction model considering climbing features quantities.Firstly,the improved bump detection technology is used to identify and divide the power fluctuation period.Secondly,this paper comprehensively considers the power fluctuation characteristics of different typical meteorological days to classify the meteorological data.Thirdly,considering the power volatility of wind power,this paper proposes a hybrid prediction model,which adopts a LightGBM-LSTM point prediction model in the power discontinuous fluctuation section and an RF-LSTM interval prediction model in the power continuous fluctuation period,and obtains a good prediction effect.Finally,the data of an offshore wind farm in eastern China is selected for improved simulation and case analysis.The results show that compared with the traditional wind power point prediction and interval prediction methods,the prediction accuracy of the partition hybrid prediction model considering wind power climbing and meteorological daily classification proposed in this paper is significantly improved.
climbing feature quantitieswind power climbing eventsbump event detectionmeteorological day classificationcombination modelpartition forecasting
NETL
NSTL
万方数据
