Reliability Evaluation Of Wind Turbine Based On Fault Correlation Bayesian Network

With the introduction of the National Carbon Neutral Strategy,wind power generation on power system has entered rapid-developing stage on clean energy utilization.The probability of equipment failures,which are usually independent and sometimes related failures,has been increasing so far these years as cumulative time accumulates and finally affects the reliability of the whole wind turbine.The large-scale mechanical equipment wind turbine is affected by both complex and changeable factors such as working environment and task conditions during operation.For the differences and dependency in mechanisms,each subsystem exerts different influences on the reliability of the whole machine and the correlation among faults varies.Therefore,based on the evaluation of the weight of the wind turbine subsystem,considering the fault related characteristics between the subsystems,the reliability evaluation of the wind turbine system is carried out by Copula Bayesian network model.Firstly,in order to solve the problem that traditional methods can’t take all the factors in the weight distribution of wind turbine into account,a combined weight method mixing fuzzy analytic hierarchy process,entropy weight method and historical fault maintenance data method is proposed,so as to obtain the weight of each subsystem of wind turbine considering expert opinions and historical fault maintenance data factors.Thus,the weight of each subsystem of the wind turbine is obtained under the comprehensive consideration of expert opinions and fault data factors.Secondly,the correlation between the two subsystems is analyzed with the method of Inference Functions for Margins to fit the parameters of the Copula function.At the same time,the subsystems with a strong correlation according to the parameter values of the Copula function is determined,and the minimum Euclidean distance method to obtain the optimal Copula Functions are used to describe the correlation between them.Thirdly,the actual wind turbine fault maintenance data is analyzed,then assign the root node with the failure probability and think over the weight based on the Copula Bayesian network model,and finally the reliability of wind turbine is obtained.The boundary theory shows that the model can describe the reliability of considering the variables’ weight and the existence of correlation very well.This paper proposes a Copula Bayesian network model with weights,which solves the problems of incomplete consideration of weight distribution factors and poor efficiency of related reliability analysis to a very extent.And thus provides a theoretical basis for wind turbine reliability research considering the importance and relevance of subsystems.