Font Size: a A A

Research On Fault Diagnosis Method Of Control System Based On Bayesian Network

Posted on:2019-07-20Degree:MasterType:Thesis
Country:ChinaCandidate:X W ZhuFull Text:PDF
GTID:2428330548989286Subject:System analysis, operations and control
Abstract/Summary:PDF Full Text Request
Control system fault diagnosis technology has been a hot research topic in the field of industrial process control.Although some achievements have been made,it still has great limitations to modern large-scale complex control systems.With the development of artificial intelligence technology,intelligent fault diagnosis technology has gradually become the focus of research.Based on the Bayesian formula,Bayesian network expresses the relationship among various variables graphically and shows great advantages in dealing with uncertainties.The application of Bayesian network in industrial control sys tem fault diagnosis can give full play to the advantage of Bayesian network in dealing with complex system and fault diagnosis system with uncertain factors.This article mainly studies from the following aspects:Firstly,a method to establish the fault diagnosis subsystem of unknown input observer is proposed,and the residual of the observer subsystem is fused into the diagnostic Bayesian network.The conditional probability table of each node is obtained through the established Bayesian network,and then the fault probability of each component is deduced by using the posterior probability calculation method of the residual node so as to realize the fault diagnosis and separation of the control system.Finally,the four-tank as an example to verify the feasibility and effectiveness of the method.Secondly,as the complexity of modern control systems becomes more and more strong,it becomes very difficult to establish a system observer.On this basis,AHP-K2 algorithm is proposed.The algorithm is a pure data-based approach and does not require the establishment of a mathematical model of the system.Firstly,the order of variables is determined by AHP,and a Bayesian network model is generated.Then the analysis of the failure analysis indexes such as aggregation coefficient,average degree and average distance of the network is used to identify whether the system operating status changes or not.The typical industrial control system(TE)process is simulated by SIMULINK,the TE process model is studied,and the method proposed in this paper is validated by TE process.Finally,the method is applied to the four-tank fault diagnosis,comparing the advantages and disadvantages of the two methods.
Keywords/Search Tags:fault diagnosis, Bayesian network, observer, residuals, AHP
PDF Full Text Request
Related items