Research On Data Analysis And Power Prediction Of Wind Turbines

Wind turbines through wind driven to make the rotation of the generator so as to achieve the purpose of power generation,because of the wind turbine drive source-the wind is uncontrollable,so you need to use some methods to control the fan to meet the requirements of power system operation,in order to control the fan capture maximum wind energy,first you need to the wind of the information collection and analysis,and control of a fan power is the power curve.Power curve is the wind turbine output power and the relationship of the third power of the wind speed curve,as the change of wind speed,the output power is also a certain trend change,when the wind speed increases to a certain value of rating changes,no longer under the actual working condition,the power curve may be higher or lower than rating,at this point is the need to control the fan to make it meet the maximum rated power output or protect the fan will not overload operation,when the fan failure will appear abnormal points in the power curve coordinate system,on the analysis of the abnormal points available to fan the cause of the problem,so the power curve to evaluate performance of the fan.The ideal power curve provided by the fan manufacturer does not take into account the actual environmental factors of the fan,so it is very important to obtain the power curve that best reflects the actual working conditions for the efficient and safe operation of the fan.There are many factors that affect the power curve.In this paper,the wind field data collection and the factors affecting the output power of wind turbine are studied.In order to improve the accuracy of subsequent fitting,the original wind field data collected should be processed accordingly.Firstly,the collected data are modified and cleaned.In this paper,a grid method is proposed to optimize the laida criterion to detect and remove the abnormal points in the scatter diagram,and finally obtain the pure data that can be fitted.This article USES the power of the most commonly used three kinds of traditional parameter method to modeling of power characteristic curve,but traditional curve modeling method is only in view of the existing data processing of fitting,not considering the influence on the output power of wind also fail to wind speed wind power prediction power relations of learning,therefore,need the power curve fitting out the more accord with the actual working condition,this article through to three methods of neural network,BP neural network,GA to optimize the BP neural network,and limit of machine learning algorithm of power curve modeling results of the comparative analysis,finally put forward a kind of extreme learning machine curve fitting based on GA optimization modeling method,This method combines the advantages of genetic algorithm and extreme learning machine,and the fitting effect is better.Before power prediction,first for wind speed forecasting,this paper proposes a EMD optimize Elman modeling algorithm,with the Elman and EMD-BP prediction algorithm to predict the results of comparative analysis,the prediction of wind speed corresponding to the four kinds of modeling method for fitting the power characteristic curve of the wind power prediction,by the wind generator power prediction result is closest to the actual operation of the wind-power curve.