Research On The Method Of Short-term Wind Power Forecastion

Wind power generation as a clean,efficient renewable energy,vigorously develop wind power generation can effectively alleviate the problems of energy depletion and environmental pollution,but due to the randomness,volatility and intermittency of wind power generation,The forecasting accuracy is higher than that of load forecasting,and large-scale wind farm grid connection will bring a lot of adverse effects to the safe operation of the system.Therefore,this paper focuses on short-term wind power prediction,the main research contents are as follows:The historical data of wind farm are preprocessed,the abandoned wind data and invalid data are removed from the historical time series,and the missing data are corrected by fuzzy reasoning.Using Pearson moment correlation coefficient to analyze the correlation between meteorological factors(wind speed,pressure,wind direction,temperature,humidity)and wind power,wind speed and wind direction are selected as the dominant factors affecting wind power.It lays a foundation for wind power prediction.Numerical weather forecast and time series are used to predict wind speed and direction.When we use the meteorological data provided by numerical weather forecast,we need to correct the wind speed data.In this paper,we use BP neural network to modify the meteorological information provided by numerical weather forecast.When using time series method to predict wind speed and wind direction data,we should first analyze the smoothness of wind speed and wind direction time series,and then use ARIMA model to predict wind speed and wind direction data.Considering the different influence of wind speed and wind direction on wind power,the idea of hierarchical clustering is adopted and the fuzzy C-means clustering algorithm is applied to the historical data of wind farm.The data of wind speed and wind direction obtained from time series and modified data are clustered,and the neural networks are trained for each category of historical data.The clustering results of weather forecast based on time series and modified data are matched with those of historical data,and the wind speed and wind direction data obtained by two different forecasting methods are brought into the corresponding neural network to predict wind power.Finally,taking the operation data of a wind farm in Taonan,Jilin Province,as an analysis object,And the prediction error of the two different methods is compared.The simulation results show that the power prediction results based on the modified numerical weather forecast are closer to the actual output results of wind farms than the power prediction results based on time series.