Early Warning Of Flue Gas Energy Recovery System With Missing Data

Flue gas energy recovery system(FS),engaged in offerring wind for carbon burning,is essential for fluid catalytic cracking,which stops when FS fails.Components interact mutually in the process of failure evolution and propagation.Although failures can be detected and repaired through traditional fault diagnosis methods,such as condition monitering,nondestructive testing,correlationships between components are ignored.Thus root causes and potential risks still exist.Also,in real project,data missing is common,however,these data are fundamental input for fault diagnosis methods.Thus,an early warning method of FS under data-missing condition is presented and a degradation trend prediction model of the main components of FS is built.On the basis of degradation prediction,optimized maintenance strategies are given.Main works are listed as follows:(1)To deal with the complex structure,various parameters and mutual effects between components of FS,an early warning model based on failure mode and effects analysis and dynamic Bayesian network is built after analyzing interrelationships between componets.(2)After studying how protective layers effect componets’ degradation,an early warning model considering protective layers’ effects of FS degradation failures is built on the basis of dynamic Bayesian networks.(3)To give certain maintenance suggestions after fault prediction,a maintenance strategies optimization method is presented,in which fault prediction,maintenance costs and effects are considered.(4)Considering traditional fault diagnosis methods are not able to deal with data missing conditions,a data implementation method considering time factor is presented to full fill missing data.