Development Of A Numerical Prediction System For Improved Wind Forecasting At Typical Wind Farms

Advances in numerical models make it possible to forecast wind speeds in wind farms.However,there are still great challenges for accurate forecasting of wind speeds under complex terrain conditions.although the forecast of local wind farm wind speeds has great socioeconomic impacts.Based on advanced numerical forecasting theories and methods,this study analyzes the basic characteristics of wind speeds in wind farms in Hexi,Gansu,based on the characteristics of complex terrain,various types of underlying surfaces,and sparse observation sites in Hexi,Gansu.The numerical simulations of wind speed proved the applicability of the numerical model parameterization scheme;carried out targeted numerical model improvement work;Then,the Gansu Renewable Energy High-resolution numerical prediction System?GREHS?has been built and used for business.After a period of stable operation,a bias correction module based on mathematical statistics was developed.The inspection and validation show that the wind speed forecast of The Green Sea system in actual business can well meet the service demand.The research in this paper mainly includes the following four aspects:1.Wind patterns and their transitions in Hexi area are explored,diurnal variation characteristics of wind pattern is revealed,and the specific relevance for the conversion of major wind types is identified.It is found that the wind field in Jiuquan region of Hexi can be divided into 5 types:"consistently weak westerly type?type I?","west to south wind type?type II?",and"valley strengthened westerly type"?Type III?,"Northern Strong Westerly Type?Type IV?",and"Weak Easterly Type?Type V?",of which Type I mainly occurs at night time and Type II mainly occurs during the day time;according to the evolution characteristics of the circulation field,different wind field types cannot be arbitrarily converted.There are transition types in the conversion of wind field types.For example,"weak westerly wind types"cannot be directly converted into"weak east wind types",and they must first undergo a transition from"west wind to south wind types".2.Analyzed the different of wind bias in different simulation results driven by different initial-boundary field,the applicability of boundary layer and radiation scheme over Hexi area are also studied.And found that the simulation bias of FNL data as the initial boundary value driving field is smaller than that of ERA5 data.The YSU boundary layer scheme has the best effect on wind speed simulation and the simulation result is more sensitive to the radiation scheme.The WRF?The Weather Research and Forecasting Model?model was used to study the wind speed forecast under different initial fields and different parameterization schemes in the Hexi region of Gansu,and compared with the observation data of regional stations and wind measurement towers.GFS?Global Forecast System?data is better as the initial field forecast result;It can be seen from the comparison results of the boundary layer schemes that the YSU scheme has a better simulation effect in northwest China in wind speed and temperature;And the radiation parameterization scheme has a great impact on the simulation wind speed bias.3.The numerical model of the wind field was improved from the three aspects of vertical coordinate system adaptability,assimilated ECMWF data and model result correction method.It was found that the mixed vertical coordinate system had a certain weakening effect on the false gravity wave.The wind speed forecast ability was improved by assimilating the ECMWF data.And,the proposed fast cycle correction method has greatly improved the prediction of wind speed.Considering the influence of the complex terrain on the wind speed forecast in the Hexi area of Gansu,a mixed?-p vertical coordinate system is used in the WRF model.This study finds that the wind speed forecast is significantly improved after reducing false gravity waves.On the other hand,in order to solve the problem of sparse observation sites in Hexi region in Gansu,This study uses the European Centre for Medium-Range Weather Forecast?ECMWF?high-resolution data to study the effect of data assimilation on wind speed prediction under different observation densities.The results show that after assimilating wind speed.The forecast results have improved significantly.Assimilation of different meteorological elements is more significant than assimilation of different data densities.At the same time,based on historical data,a fast-cycle correction method using data mean,variance,and trend is developed in this paper.The method is used to statistically correct the model forecast wind speed.The corrected result is significantly better than the model direct forecast result.4.A wind power forecasting system has been built in Hexi area and it is used to predict wind speed.The result show that the wind speed bias is small,which meets the business requirement.Based on the above researches,and combined with the improvements in the B-matrix,Landuse,and vertical layer encryption,this study establishes a green energy system for the Hexi region of Gansu.The numerical forecasting system is used to carry out batch forecasting research.The forecast objects include five wind farms from January to May 2019.Using batch technology to further improve the correction scheme,the results show that the forecast error of most stations?time?is less than 2 m·s^-1,which meets the business needs.The results of this study can be generalized to numerical forecasting of wind farms under complex terrain.The GREHS from this study is the first wind energy numerical model forecasting system suitable for complex terrain conditions in Northwest China.