Premium
Real‐time diagnostics of the olefin production process
Author(s) -
Osipenko Uliana,
Rusinov Leon
Publication year - 2019
Publication title -
journal of chemometrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.47
H-Index - 92
eISSN - 1099-128X
pISSN - 0886-9383
DOI - 10.1002/cem.3118
Subject(s) - subspace topology , process (computing) , principal component analysis , residual , computer science , frame (networking) , task (project management) , production (economics) , sensitivity (control systems) , artificial intelligence , pattern recognition (psychology) , data mining , algorithm , engineering , economics , macroeconomics , operating system , telecommunications , systems engineering , electronic engineering
The olefin production process is explosive and fire hazardous. Thus, the development of a diagnostic system for earlier identifying abnormal situations and determining their causes is an important task. Analysis of the process is executed to elaborate a diagnostic model. Possible abnormal situations, their location, and symptoms are determined. It is proposed to realize process monitoring with the aid of widely used PCA. Monitoring of the process state is carried out according to the PCA model by the control of statistics T 2 and Q in the principal component subspace and residual subspace, respectively. To determine the causes of abnormal situations, it is proposed to use a combined two‐level frame production diagnostic model. The upper level of the model is formed by the network of root frames, and the lower level contains daughter frames with production rules that describe abnormal situations. The working capacity of the model is shown with the help of a specialized expert shell. The possibility of using a diagnostic system to monitor the activity of catalyst and maintain roughly constant yield of olefins during catalyst coking is shown.