Research On Short-term Wind Power Prediction Based On Meteorological Model Correction

Wind power generation has the advantages of clean and pollution-free,low power generation cost and strong sustainability,which makes it one of the most potential renewable energy power generation technologies.With the rapid expansion of the scale of wind power grid connection,the uncertainty of wind power generation has caused many adverse effects on power quality,power grid dispatching,safe and stable operation of power system and so on.However,accurate wind power prediction can effectively avoid the problems caused by wind power generation.Therefore,there is an urgent need to improve the prediction accuracy of short-term wind power.So,the main research work of this paper is as follows:When establishing the power forecasting model,considering the meteorological factors with weak correlation will increase the training difficulty of the model,thus affecting the prediction accuracy of wind power.For this reason,a correlation analysis method based on mutual information and information entropy theory is used in this paper.On the basis of combining with the traditional correlation analysis method,this method calculates the correlation value accurately.The correlation between multi-dimensional meteorological factors and wind power is quantified to avoid the influence of subjective selection on the accuracy of wind power prediction.Based on the analysis of the wind farm of Da-Lin-Tai,it is shown that four meteorological factors which have great influence on the output of the target wind field are obtained by this method.In addition,the dimensionality reduction of multi-dimensional meteorological factors is realized,and the accuracy of wind power prediction is improved finally.Numerical weather forecast(NWP)cannot provide high-precision weather forecast for each wind field,there will be a large error between the weather forecast data and the actual data,which will affect the prediction accuracy of wind power.For this reason,a meteorological data correction method based on numerical weather forecast is proposed in this paper.In addition,a meteorological correction model based on genetic algorithm(GA)to optimize long-term and short-term memory(LSTM)network is established.First of all,the distribution of the error probability density is obtained based on the prediction error of NWP,which increases the correction accuracy of the meteorological model.Secondly,the structure of GA and LSTM network is described in detail,and a method of optimizing LSTM neural network based on GA is proposed.Finally,a meteorological model based on the optimized LSTM network is established to modify the meteorological data needed to predict wind power.Among them,the measured meteorological data at t-1 time and NWP at t time are input,and the measured meteorological data at t time are output.Based on the analysis of an example,it is shown that for zonal wind speed,radial wind speed and relative humidity variables,the prediction accuracy can be improved by about 5% after correction by the meteorological model,which verifies the rationality and effectiveness of the meteorological modified model.The randomness and intermittence of fan output affect the safe and stable operation of modern power system.For this reason,a short-term wind power forecasting method based on meteorological model modification is proposed in this paper.First of all,the data of the original wind power is preprocessed based on wavelet analysis and threshold denoising to avoid the impact of wind farm acquisition environment and devices on power prediction accuracy.Secondly,a short-term prediction model of wind power based on GA optimized LSTM network is established to predict the output power of wind field at t time.Among them,the meteorological forecast data corrected by the meteorological model at t time and the actual wind power after noise reduction at t-1 time are inputs.Finally,the prediction error is analyzed to establish a theoretical basis for further improving the accuracy of wind power prediction.The analysis of an example shows that the hybrid algorithm proposed in this paper can improve the accuracy of wind power prediction and is easy to be realized in engineering.