Short-Term Wind Power Forecasting For Wind Farms

With the development of society and the advancement of science and technology,the world’s demand for energy is increasing,and traditional fossil energy can no longer meet people’s current demand for energy.In order to solve the world’s energy problems,people have developed wind energy,tidal energy,solar energy and other renewable energy sources,wind energy has a relatively high proportion of the utilization of many renewable energy sources due to its short infrastructure period,flexible installation scale,and low operation and maintenance costs.However,the instability of wind energy also brings troubles to the large-scale grid connection of power systems.In order for the power grid to operate safely and stably,it is necessary to predict the power of wind power.In order to arrange the power generation plan more rationally,this paper makes short-term deterministic forecasts and interval forecasts for wind power.This article first outlines some of the problems in wind power forecasting,and then begins to calculate and select the input parameters of the wind power forecasting model.The gray correlation method is used to calculate the gray correlation between each parameter in wind energy and the wind power of the wind farm.Among them,the parameters with the highest gray correlation value are used as the final input of the prediction model,and finally the wind speed,wind direction and temperature are determined as the input of the wind power deterministic prediction model.Then use the Laida criterion to calculate and filter the original data of each parameter,eliminate the wrong data in the original sequence,and use the linear regression algorithm to calculate and correct the deleted data to ensure the integrity and accuracy of the original sequence of each parameter.The influence of the dimension of the parameter on the prediction result is normalized and input into the prediction model.Aiming at the problem of wind power forecasting,this paper divides it into point forecasting and interval forecasting based on point forecasting.When forecasting wind power points,firstly use gray neural network,radial basis neural network and cyclic neural network to model wind power power point predictions,and then use average absolute percentage error and root mean square error as the model’s error evaluation indicators,Compare the performance of the three neural network models,analyze the final experimental results,and successfully verify that the gray neural network has the best performance among the three prediction models.Then use dragonfly algorithm,bacterial foraging algorithm,and firefly algorithm to optimize and improve the gray neural network model,and use Matlab to simulate the model under the same input and data conditions.The average absolute percentage error and root mean square error are also used.As the error evaluation index of the optimized gray neural network model,the three optimized models and the gray neural network are compared,and finally it is concluded that the three algorithms have optimized the prediction performance of the gray neural network.Among them,the dragonfly algorithm The optimization effect is the best,successfully verifying the validity and superiority of the DA-GNN model in wind power prediction.Finally,based on the deterministic prediction of the DA-GNN model,the shortterm interval prediction of wind power is carried out,and the quantile regression algorithm is used as the experimental model of interval prediction.Take the wind speed,wind direction,temperature and the prediction error of the DA-GNN model as the input of the interval prediction model,and the wind power as the output.Use Matlab to carry out simulation experiments of the model at the confidence level of 90%and 85%respectively,and use the prediction interval The coverage rate and the normalized average width of the prediction interval are used as the evaluation index of the model to analyze the model.The prediction results and error analysis prove the feasibility and effectiveness of the combined model in the prediction of wind power interval.