Research Of Short-term Wind Power Prediction And Its Application In Gansu Grid

In today's background of traditional petrochemical energy draining and have large damage to the environment,new energy sources such as wind power is paid attention to by the people day by day.Because of the big fluctuation and strong intermittent of the wind,it is necessary to predict the wind power to facilitate the management and scheduling.This paper firstly discusses the power characteristics of wind farms,point out the difference between wind power actual operation curve and the power characteristic curve,denied the power prediction method relied on power characteristic curve.Then introduces the core technology such as data preprocessing,input variables selection,intelligent algorithm parameters optimization involved in the wind power prediction,then based on the characteristic of wind farm,mis paper proposes a short-term wind power prediction method based on least squares support vector machine(LSSVM).In order to eliminate the prediction error of wind farms because of different fan type and installation location,the grouping prediction method was adopted.Finally,through the establishment of forecasting model,and uses the typical actual operation data of a wind field with four months of spring,summer,autumn and winter to prove the feasibility and effectiveness of the method proposed in this paper.The grouping prediction error and wind field prediction error are reaching or lower than the technical requirement of the wind power prediction.By grouping prediction method,compared with a single set of fan,the prediction error of the prediction results after superposition also have varying degrees of lower.According to the method developed wind power short-term wind power prediction system,the actual field operating data shows that the method and the system has good prediction effect,can satisfy the scheduling needs of dispatching department to supervise and control the wind farms' wind power.