Research On Short-term Power Forecasting Method Of Wind Turbine Based On Neural Network

The sound development of new energy such as wind power can effectively help my country occupy an important position in the future energy market.With the improvement of the grid-connected capacity of wind turbines and the increase of wind energy market share,effective storage and utilization of electric energy generated by wind turbines is an effective means to improve the utilization rate of wind energy.In wind turbines,the wind energy generated in the future can be effectively predicted and the electric energy generated in advance Preparing energy storage space is a good way to improve the utilization rate of wind energy.In the current wind power forecasting,there are problems such as low efficiency of wind power data preprocessing,difficulty in effectively identifying abnormal data,and low accuracy of wind power forecasting.Aiming at the problem that it is difficult to identify the abnormal data of the wind power data set,this paper designs an improved Isolation-Forest algorithm based on the influence factors of characteristic anomalies,and builds a back propagation neural network model for wind power prediction.This paper uses the Mean Square Error curve between the wind power prediction value and the real value to compare the current popular anomaly identification algorithms such as the quartile method and the density-based DBS CAN clustering algorithm.The experimental results verify the effectiveness and superiority of the proposed Isolation-Forest algorithm based on the characteristic anomaly influencing factors in the identification of wind power anomaly data.It provides a new solution for the study of abnormal data identification in wind power datasets.Aiming at the common problems in wind power forecasting such as low precision of wind power forecasting and slow error convergence,this paper designs a back propagation neural network model based on the improved Adam algorithm as a forecasting model of wind power.In this paper,the convergence characteristics of the error curve and the minimum value of the error are the research characteristics,and the performances of the traditional back propagation neural network and the back propagation neural network model based on the Adam algorithm are compared.The experimental results have verified that the back propagation neural network model based on the improved Adam algorithm has the advantages of faster error convergence speed and stable convergence.It has a good application prospect in the project of wind power forecasting.