Abstract
AbstractInformation delay problem connected with the nature of flowing information in process industry is studied. An information synchronization technique is proposed using time-delayed mutual information, to determine the information's directionality and time delays between process variables. Signed Digraph model is derived to represent the information's directionality and lags. A principle component analysis (PCA) based fault prognosis method is finally proposed to achieve early detection of plant-wide process abnormalities. The proposed method was applied to an air separation unit and achieved satisfying results in predicting the frequently occurred nitrogen-block faults.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
