Research On Short Term Power Prediction Algorithm Based On Dynamic Adjustment Method Of Wind Farm Cluster

At present,China’s energy consumption structure is accelerating the transition to clean and low-carbon.In order to balance energy security and low-carbon transformation,it is necessary to vigorously develop and utilize non-fossil energy such as wind energy,solar energy,and biomass energy.Among them,wind energy has huge reserves and is clean and renewable.and other advantages have been developed rapidly.The fluctuation and randomness of natural wind determine the uncertainty of wind power.With a large number of wind turbines being connected to the grid,it affects the reliability and safety of power system operation.Therefore,accurate prediction of wind power can ensure the stable operation of wind farms.It can reduce the dispatching burden of the power grid system.In this paper,a prediction method of wind farm cluster power is proposed.The main research work is as follows:The theoretical basis of the maximum information coefficient(MIC)and the calculation process of MIC are analyzed,and the MIC is compared with other correlation measurement methods,and the better stability and applicability of MIC are verified by typical signal simulation.Conduct experiments based on the measured SCADA data,quantitatively analyze the influence of various state parameters on the fan output,and provide a basis for the selection of input variables for the prediction model.The fluctuation characteristics of wind turbine output power caused by the uncertainty of natural wind are studied.Variational modal decomposition(VMD)is introduced and the simulated composite signal is used to test the anti-modal aliasing characteristics of VMD compared with empirical mode decomposition(EMD).;Using the measured data of wind farms to conduct experiments to verify the applicability of VMD and the proposed decomposition parameter determination method based on correlation coefficients in wind power.Considering the lack of information caused by VMD decomposition and the influence of environmental factors on modal components,a combined prediction algorithm for wind turbine power based on variational modal decomposition(VMD)and maximum information coefficient(MIC)is designed.Aiming at the randomness and volatility of wind power time series,VMD is used to decompose the original wind power series into modal components with different fluctuation characteristics;The components are selected for feature selection;based on the induced ordered weighted average(IOWA)operator,a combined model is established to predict sub-items,and finally the prediction results of each modal component are superimposed to obtain the final predicted value.Experiments are carried out based on the measured data of the wind farm,and the results show that the proposed combined forecasting model can effectively improve the forecasting accuracy.On the basis of realizing the combined forecast of a single wind turbine,a shortterm forecasting algorithm of wind power based on the dynamic adjustment method of wind farm clusters is proposed.Firstly,the time scale to be divided is determined by analyzing the collected data of the wind farm;the wind turbine cluster is dynamically divided according to the distribution of wind farm fans and the difference of wind resources;the dynamic time warping(DTW)algorithm is used for each cluster to select The representative wind turbine of the cluster uses the proposed combined model to predict the representative wind turbine;while the sub-cluster uses the LSTM model to predict the power value of the cluster,and superimposes all the cluster predicted values at this time scale to obtain the total power predicted value of the wind farm;Integrate the forecast values of multiple time scales to obtain the wind power forecast results of the entire wind farm for a long period of time.Experiments are carried out with the measured data of the wind farm to verify the validity of the proposed wind farm prediction model.