Research On The Method Of Non-complete Lifetime Data Analysis Based On Failure Mode

Lifetime data analysis plays an important role in the reliability data analysis of the equipment.The development of reliability data analysis in domestic is relatively late,and it is most concentrated in the military and aerospace and other high risk industries.In recent years,the accident caused by the unreliable petrochemical equipment happened quite frequently,so it is necessary for petrochemical equipment to collect and analyzing reliability data to ensure normal operation of equipment.In addition,it is necessary to explore the reliability data analysis method of petrochemical equipment because of the time of collecting petrochemical equipment reliability data of is short and the shortage of systematic management of data.In view of the above situation,the main research content of this paper is:(1)Basic terms of life data analysis were overviewed briefly and the significance of data collection and classification of equipment life was clarified.In the meantime,based on the failure mode and failure cause of equipment,the samples of life data were classified.Ulteriorly,the common software of reliability data analysis and the commonly used reliability database of petrochemical equipment were introduced.The petrochemical equipment reliability data management system,an application of Access and Excel as the database store background and the database query front desk relatively in the system,was preparaed.(2)For the homogeneous samples,optiming right censored data maximum likelihood estimation method using Monte Carlo method led to the small mean square error in parameter estimation and a critical statistic value was calculated to choose the estimation method.In addition,the iterative formula of MLE of left censored and interval censored was deduced and it was been programmed.Finally,the parameter estimation mode of homogeneous samples was be established and it applied in safety valve reliability data come from a certain petrochemical enterprises.(3)Regarded the safety valve lifetime data come from catalytic cracking unit in a certain etrochemical enterprises as homologous multiple samples,then mixed distribution model was used to analyze the homologous multiple samples and the analytic hierarchy process was applied to determine the mixing parameters of mixed distribution.The comparison between Mixed distribution model and Mixed samples distribution model was be done.The comparison results show that the Mixed distribution model is more effective than the usual mixed sample distribution model in the practical lifetime data analysis.In view of the lifetime data from different failure modes under different environments,it is considered as heterologous multiple samples data and the competitive failure model was used to analyze the sample.(4)Aimed at the situation of no failure data,OREDA database was applied as the historical data of the equipment and the centrifugal pump system was taken as an example.Firstly,after a lot of calculation and complex finishing,failure data of the centrifugal pump system was arranged for each unit fault number based on fault fault severity level.Using Weibull distribution to fit the unit distribution.Combining the reliability function and mean time to failure formula,the unknown parameters of each fault severity level can be calculated and using competition failure model calculated the reliability function for each unit.Drawing the reliability block diagram of centrifugal pump system,finding the reliability function of the centrifugal pump system,and using Origin software to plot the graph of reliability function and the graph of probability density function of centrifugal pump system and its units,then analysing the regularity of reliability of centrifugal pump system and its units.The analysis results show that it is necessary to strengthen the inspection and maintenance work of the centrifugal pump system in the first 1000 hours of its life cycle,and to ensure the good operation of the system.