Reliability Analysis For Wind Turbines Based On Bayes Monte Carlo Method

The consumption of fossil fuel has raised widely concerns over energy supplement in recent years.With the increasing of quantity and operation time,wind turbines are subjected to several kinds of failures.Wind turbine is large-scale outdoor running equipment,the working environment of which is very harsh and far away from city.Whenever it failed,it cost much to repair.It is essential to model the reliability of wind turbines,which should be in accordance with the actual operation.In order to satisfy the aforementioned requirement,this paper takes the whole wind turbine as research target,ignoring the inner structure of it,and model the reliability variance for it.In this paper,a wind farm which is located at the north west of China has been taken into consideration.It is about three years of failure record has been used to model the reliability of wind turbine.Bayes method is applied to estimate the model parameters,whereas Monte Carlo simulation is used to enlarge the data sample and Gibbs sampling is selected to figure out the posterior distribution.These three different distributions,Weibull,Exponential,Lognormal,has been compared with each other.The reliability obtained by these three distributions is almost the same.It turns out that when the operation time is over 5 years,the reliability decrease to 0.6,and then the decreasing speed slow down.But the failure rate calculate by these three distributions is extremely different.Given the Bathtub curve theory,the Weibull model obtained by Bayes method is determined as the reliability model for wind turbines in final.The method which is used to model the reliability in this paper can be applied to model the reliability for the other wind turbines around China.And the result of the reliability provides useful messages for planning preventative maintenance.