Study On Wind Turbine Gearboxes Health Statues Valuation Based On Healthy Samples And Trend Prediction

China's wind industry has experienced three stages which are technology importation,digestion and absorption,self-innovation,so far,the total installed capacity has ranks first in the world.Bad environment,complex running status and many other uncertainties caused high maintenance costs.In recent years,with the research and application in wind power industry of mechanical enginee ring,electric engineering,artificial intelligence etc.our wind turbine technology has made substantial devevelopment.Through condition monitoring on blower,we can find and solve the problems earlier.During the whole life of blower,people always seek a goal of reduce the operation and maintenance cost,meantime,improve the generating efficiency.So the effect of state recognition technology are becoming more and more obvious.Statistics show that the gearbox failure caused wind turbines downtime and maintenance costs much higher than other failures.There has been a lot of research for gearbox health assessment technology,but the current status assessment method have some shortcomings.First of all,due to the lack of wind turbine gearbox fault samples,and there is not enough research to prove that different wind turbines between the gearbox operating characteristics of the same;Second,evaluation based on the current unit operating data can not fully reflect the health status of the unit,for example,when the deterioration trend is obvious and the eigenvalue is at the safe level,this state characteristic needs to be highlighted.We propose a method to evaluate the health status of gearboxes of wind turbines by health samples and trend forecasting based on the previous research in this paper.The method includes the following steps: Firstly,use data mining technology to mine the large amount of health wind turbines data.Then built the health parameter model by training the improved BP neural network based on the genetic algorithm.The health parameter model is established,and the residuals of the characteristic parameters are obtained by comparing the experimental values with the model.Finally,the residual value is evaluated by fuzzy comprehensive evaluation theory to evaluate the gearbox.This method is characterized by fitting the multi-parameter health feature surface from a large number of wind turbine condition monitoring data,combined with the fuzzy comprehensive evaluation method with trend prediction,to evaluate the health status of the wind turbine generator gearbox and whole machine.