Research On Short-term Wind Power Prediction Method Of Wind Farm Cluster

Nowadays,China's economy is developing rapidly,and the massive consumption of fossil energy has caused serious pollution to the ecological environment.The development of wind energy plays an irreplaceable role in optimizing the production capacity structure and adhering to the implementation of sustainable development strategies.However,wind power has the characteristics of gap,randomness and volatility,so it is very difficult to use wind energy.The main influence factor of wind power is wind speed,and the output power of wind farms will also fluctuate,which will have a certain impact on the safe and stable operation of the power grid.The impact of output volatility on the power grid improves economic efficiency.With the rapid development of wind power,wind farms in many regions are clustered and scaled.In terms of dispatching,the staff pays more attention to the power of the entire wind farm cluster area.necessary.The accurate prediction of the wind power in the wind farm cluster area is more conducive to the scheduling and operation mode.Therefore,this paper proposes a wind farm cluster output prediction method.Firstly,this paper describes the background and significance of power forecasting of wind farms,introduces the development status of power forecasting of single wind farms and wind farms at home and abroad,and describes the research scope and classification of wind power forecasting.It mainly analyzes the characteristics of wind power output,and qualitatively and quantitatively analyzes the factors that affect the output of wind turbines,which provides a scientific basis for the selection of input variables for wind power prediction models.A single wind farm prediction model based on EMD-LSTM is established.First,the relevant data is preprocessed to obtain the ideal input sequence.The LSTM network model is used to first modify the numerical weather forecast(NWP)wind speed to obtain the revised forecast day NWP wind speed sequence Closer to the actual wind speed.Then,the empirical mode decomposition(EMD)is used to decompose the wind power data sequence into data components of different scales,and then the LSTM long-short-term memory network is used to construct the respective intrinsic modal function(IMF)components and residual RES separately Mode,the results of modeling predictions of each component and margin are added together as the final result of wind power prediction.Based on a single wind farm prediction model,a wind farm cluster output prediction method is proposed,which includes dividing sub-regions,selecting representative wind farms,representative wind farm output prediction,sub-region output prediction,and sub-region prediction result summation.The sub-regions are divided according to the geographical location of the wind farms in the cluster and the characteristics of the wind energy in the region.The representative wind farms are selected according to the correlation coefficients and relevant factors of the output of each wind farm and sub-region.For the prediction model,the LSTM model is still used to predict the total output power of the sub-region.Finally,a wind farm cluster in an area of Inner Mongolia was used for experiment and error analysis to verify the effectiveness of the method described in this paper.