Font Size: a A A

Fault Diagnosis Of Gearbox Based On Improved Bayesian Network Model

Posted on:2020-06-16Degree:MasterType:Thesis
Country:ChinaCandidate:Y ZhangFull Text:PDF
GTID:2392330575453175Subject:Engineering
Abstract/Summary:PDF Full Text Request
With the concept of industry 4.0 and the putting forward and implementing of "2025" plan of Made in China,the industrial production of our country is more and more developing in the direction of intelligence and network.As the core component of all kinds of rotating machinery,the normal operation of gearbox has a vital impact on the vigorous development of the whole industry.Because it is a sealed box structure,difficult to maintain,time-consuming and costly,its normal operation and accurate diagnosis are directly related to the safety and economic benefits of industrial production.In view of the fact that there are great uncertainties among the factors inducing the failure of the gearbox,this paper intends to adopt the improved Bayesian network model to diagnose the gearbox accurately.Firstly,using the method of fault tree analysis,the uncertain factors causing gear box failure are analyzed,and the failure fault tree is established.According to the expert prior knowledge obtained from various ways,the failure fault tree is transformed into a Bayesian network model which combines expert prior knowledge.The model is used as the inference engine module of expert system.Then,using the Bayesian network learning toolbox Full BNT-1.04 of MATLAB,combined with the training data samples under normal conditions,the structure learning and parameter learning of the primary Bayesian network model established by combining the prior knowledge of experts are carried out to obtain a Bayesian network model more suitable to the actual data.Finally,a gear box fault simulation experiment is designed.Firstly,the validation sample data set is input into the Bayesian network model combined with expert prior knowledge,and then the same validation sample data set is input into the Bayesian network model after structure learning and parameter learning.Comparing the fault diagnosis accuracy of the two models,it is proved that the improved Bayesian network model has better fault diagnosisaccuracy of gearbox.In summary,the Bayesian network model studied in this paper based on actual data can accurately diagnose the fault factors with great uncertainty of gearbox,and provide some help for the future research of gearbox fault diagnosis.
Keywords/Search Tags:Bayesian Network Model, expert system, fault diagnosis, Structural Learning, Parameter learning
PDF Full Text Request
Related items