Research On Improved Bayesian Network Algorithm For Fault Diagnosis System Of Cement Rotary Kiln

Cement clinker firing system is the most important part of cement production,the most critical clinker calcination physical and chemical reaction in the rotary kiln,a direct impact on the cement clinker production,quality and energy consumption.Although the failure rate is decreasing as the reliability of the equipment is improved and the production process is improved,the failure is unavoidable.The clinker firing system is not only large and complex,but also shows uncertainty and ambiguity,which makes the traditional artificial method can't meet the requirements of the diagnosis of cement clinker firing system.Based on this,the fault diagnosis model of cement clinker firing system based on improved Bayesian network is designed based on the production technology and system mechanism of cement clinker firing system.The diagnostic system includes data preprocessing,Bayesian network structure learning,Bayesian network parameter learning and Bayesian network reasoning.Through the learning and reasoning of the diagnosis model of the firing system,the fault diagnosis of the cement clinker firing system is realized.The specific contents are as follows:First of all,aiming at the existing defects of the classical Bayesian network structure learning algorithm,proposes an improved Bayesian network structure learning algorithms—Bayes Membership-threshold algorithm(BMT),the simulation results show that the BMT algorithm is based on a large number of complete data The accuracy and accuracy of the algorithm can be improved.Therefore,this paper uses the BMT algorithm as a structural learning algorithm for fault diagnosis model of cement clinker firing system.Secondly,aiming at some shortcomings of Bayesian network parameter learning algorithm,an improved Bayesian network parameter learning algorithm—Arrangement Health degree algorithm(AHD),is proposed to solve the problem of the prior distribution of classical Bayesian parameters randomness and uncertainty.Therefore,this paper chooses AHD algorithm as a parameter learning algorithm for fault diagnosis model of cement clinker firing system.Thirdly,the mechanism of the cement clinker firing system was analyzed and the process analysis was carried out.The research variables were selected and the research data were collected and quantified.In order to solve the problem of fault diagnosis of cement rotary kiln,Bayesian network is applied to fault diagnosis of cement rotary kiln.Use BMT algorithm to train the quantized data and establish fault diagnosis Bias network structure model of cement rotary kiln,using AHD algorithm parameters on the structure model of learning,finally obtained the complete cement clinker firing Bias network fault diagnosis system model.Finally,this thesis introduces the Bayesian network reasoning process,and uses the classical variable elimination(VE)method to reason the Bayesian network model of the cement rotary kiln.According to the accuracy curve obtained from the data experiment,we can see that the diagnosis model is higher Diagnostic accuracy.