Research On Fault Diagnosis Method Of Wind Turbine Transmission System Based On Information Fusion

Because the wind turbines was operated in remote and harsh working conditions,the wind turbine drive system is prone to failure under alternating loads and difficult to maintain.Therefore,it was of great significance to improve the power generation efficiency of wind turbines by analyzing the fault mechanism of wind turbine transmission system and studying effective fault diagnosis methods.However,the existed fault diagnosis methods only combine certain fault feature sets or certain pattern recognition algorithms and cloud only find the optimal fault classification method from a single point of view,which was bound to have one-sidedness and affect the fault diagnosis of the wind turbine power transmission system.This paper adopts the information fusion method to solve this problem,and studies from the aspects of feature level fusion and decision level fusion.For the modernization of rotating machinery and complicated working conditions,single or single-domain features could not fully reflect the state characteristics of wind turbine drive system,and there are redundancy and conflict problems in multi-domain features.In this paper,a fault diagnosis method for wind turbine drive system based on feature fusion of Deep Belief Network was proposed.Firstly,a multi-domain feature set that fully reflects the state characteristics of the rotating machinery was constructed.Secondly,the Restricted Boltzmann Machine in the Deep Belief Network was used to study the characteristics of multi-domain feature sets and fully explore the essential features of the data,reached the goal of fusion features and eliminating redundant and conflicting information.At last,BP neural network was applied to classify the essential features of relearning,and reversed the network weights.Through the fault diagnosis of the planetary gear box and the diagnosis experiment of the rolling bearing fault degree,it was proved that the Deep Belief Network method was useful in feature fusion and improves the accuracy of fault diagnosis.Aimed at the conflict of evidences in multi-decision results of wind turbine drive system fault diagnosis,a fault diagnosis method of wind turbine drive system based on improved D-S evidence theory was proposed.The correlation matrix between the primary diagnostic evidences was constructed by using the conflict factor in D-S evidence theory,and the reliability of each evidence was calculated from the correlation matrix.Then,according to the reliability of the evidence for similar evidences and conflict evidences classification,retained the similar evidences and modified the Basic Probability Assignment of conflict evidences to reduce the low reliability of evidence on the impact of decision-making results.Finally,the application of Dempester rule to make a decision fusion of similar evidences and modified conflict evidences.The experimental results of fault diagnosis in rolling bearing and planetary gearbox showed that the improved D-S evidence theory method can effectively solve the problem of evidence conflict in multiple decision results,reduced the uncertainty of diagnosis and improved the diagnostic rate.In this paper,the information fusion theory was applied to fault diagnosis of wind turbine transmission system.In terms of feature level fusion,the application of deep belief network method in feature fusion was studied,and the problem of information conflict and redundancy in multi-domain feature set was effectively solved;In terms of decision-level fusion,the application of improved DS evidence theory in multiple decision-making results was studied,and the problem of conflict of evidence in multi-decision results was effectively solved.Finally,the effectiveness of the proposed method is verified by experimental results.