Research On Reliability Evaluation Method Of Complex Production System In Process Industry

The complex process industry system,which is composed of mechanical and electrical equipments and production processes,is a continuous and interdependent logic production.In order to comprehensively evaluate the operation reliability,discover hidden problems in time,and accurately grasp the operation status of complex process industrial system,and developing the best solution is the main technical problem of system reliability research.In this paper,information processing,fault diagnosis,reliability judgment of single equipment,real-time reliability analysis of multi-equipment complex system,quantitative analysis and prediction of offline health status and efficiency evaluation of reliability research schemes and methods are studied.1.In order to analyze the reliability of process industry information and get the characteristic frequency of the signal,a feature extraction algorithm based on combination mode decomposition and singular value decomposition(ECMD-SVD)is proposed.The combined mode decomposition(ECMD)is composed of EMD and CEEMDAN.Firstly,the original signal is decomposed by ECMD,according to the rule based on correlation coefficient and kurtosis,the inherent mode functions(IMFs)are determined,then the first filtered signal is obtained by summing IMFs.secondly,Hankle matrix is constructed using the reconstructed signal and is decomposed by SVD,and the effective singular values are determined by Maximum difference spectrum of singular values,and then the second filtered signal is obtained by inverse operation of singular values.Thirdly,the characteristic frequency of the signal is extracted by the?method?of?DEA(?Data?envelopment?analysis).Finally,it is applied to the feature extraction of rolling bearing fault vibration signal,the experimental results show that the algorithm is effective and feasible.2.In order to diagnosis the fault,on the basis of extracting reliability information by ECMD-SVD features,the multiscale dispersion entropy(MDE)of the denoising signal are used as the feature vectors of the classification,and then Support Vector Machine(SVM)and the continuous hidden Markov mixed Gauss function(GMM-CHHM)are used to classify the fault.Two methods are named MDE-SVM and MDE-GMM-CHHM respectively.MDE-SVM simulation results show that the classification effect is significant and the signal feature extraction method based on ECMD-SVD is scientific.MDE-GMM-CHHM simulation results,based on the forward probability calibration method and the maximum likelihood probability comparison method,show that it can diagnose rolling bearing faults accurately3.In order to analyze the real-time reliability of multi-equipment complex system quantitatively,the health evaluation system of complex system running state is put forward,which takes into account factors such as product,efficiency,equipment,energy consumption and loss.The combination of reliability,mutual information entropy and behavior mode is designed to represent the system's health degree,and the feasibility of the4.scheme is verified by an example.5.In order to quantitatively evaluate and predict the health degree of a single device or complex system,an evaluation system for the health of complex systems is proposed,using AHP,matter-element information entropy and combination weighting method is proposed.The matter-element information entropy model of the single equipment and the complex system are established,after the joint weight is determined,health degree is quantitatively calculated by the method of complex matter element associated entropy.Future health degree is predicted by SVM and Least Squares Support Vector Machine(LSSVM),and the prediction model is used in the transformer and beer filling production lines,the results of the experiment are satisfactory6.In order to evaluate the accuracy and efficiency of reliability schemes and models,it is proposed that the efficiency of reliability models of mechanical and electrical equipments is evaluated by Meta.On the basis of fully studying the establishment of meta-analysis model,the judgment of heterogeneity,the combination of cumulative effect and the comprehensive analysis and evaluation of research results,Meta analysis is used to analyze the modeling effect of data-driven method,physical failure model method and the combination of the two methods on the performance degradation of similar mechanical and electrical equipment.Net meta analysis is used to evaluate the application effect of various data-driven methods on the life prediction of the same mechanical and electrical equipment.