Razvoj softverskih senzora za praćenje rada procesnih analizatora

master's thesis

• Title (english): Development of soft sensors for monitoring the operation of process analyzers
• Author: Magdalena Šimić
• Mentor: Željka Ujević Andrijić (mentor)
• Committee member: Željka Ujević Andrijić (predsjednik povjerenstva)
• Committee member: Srečko Herceg (član povjerenstva)
• Committee member: Igor Dejanović (član povjerenstva)
• Granter: University of Zagreb; Faculty of Chemical Engineering and Technology; Zagreb
• Defense date and country: 2023-07-10, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Chemical Engineering

Abstract

U ovom diplomskom radu predstavljen je razvoj dinamičkih modela softverskih senzora za praćenje rada online procesnog analizatora za mjerenje količine sumpora u proizvodu procesa hidrodesulfurizacije dizelskog goriva. Jedan od važnih koraka u razvoju softverskih senzora bio je odabir i statistička obrada kontinuirano mjerenih podataka dobivenih s postrojenja. Primjenom metoda identifikacije procesa razvijeni su dinamički polinomni linearni modeli: model konačnog impulsnog odziva, FIR (engl. finite impulse response), model temeljen na autoregresiji sa egzogenim ulazima, ARX (engl. autoregressive with exogenous inputs) te model izlazne pogreške, OE (engl. output error). Razvijeni su i dinamički nelinearni modeli: nelinearni model temeljen na autoregresiji s egzogenim ulazima, NARX (engl. nonlinear autoregressive with exogenous inputs) i Hammerstein-Wiener (HW) model. Različitim postupcima vrednovanja modela provjerena je valjanost razvijenih modela. Najmanja odstupanja modela od eksperimentalnih podataka pokazali su ARX i OE modeli te se oni mogu koristiti za kontinuiranu procjenu količine sumpora u proizvodu procesa hidrodesulfurizacije dizelskog goriva. Svrha razvijenih modela softverskih senzora je praćenje rada procesnih analizatora i potencijalno korištenje u svrhu zamjene procesnih analizatora prilikom njihovog održavanja ili dužeg kvara uređaja.

Abstract (english)

This master’s thesis demonstrates the development of dynamic models of soft sensors to monitor the operation of on-line process analyzers for measuring the sulfur content in the product of the hydrodesulfurization process of diesel fuel. The most important steps in the development of soft sensors were the selection and preprocessing of continuously measured data collected from the plant. Using process identification methods, dynamic polynomial linear models were developed: finite impulse response (FIR), autoregressive with exogenous inputs (ARX) and output error (OE) models, as well as dynamic nonlinear models: nonlinear autoregressive with exogenous inputs (NARX) and Hammerstein-Wiener (HW).
The performances of the developed models were tested by various model evaluation procedures. The models ARX and OE showed the smallest deviations from the experimental data and can be used for continuous estimation of the sulfur content in the product of the diesel fuel hydrodesulfurization process. The purpose of the developed soft sensor models is to monitor the operation of process analyzer and potentially replace it during maintenance or equipment failures.

• Keywords: empirijski matematički modeli; polinomni linearni i nelinearni dinamički modeli; softverski senzori; statistička obrada podataka; proces hidrodesulfurizacije
• Keywords (english): empirical mathematical models; polynomial linear and nonlinear dynamic models; soft sensors; statistical data processing; hydrodesulfurization process
• Language: croatian
• URN:NBN: urn:nbn:hr:149:073933
• Study programme: Title: Materials Science and Engineering - Graduate study; Study programme type: university; Study level: graduate; Academic / professional title: magistar/magistra inženjer/inženjerka kemijskog inženjerstva (magistar/magistra inženjer/inženjerka kemijskog inženjerstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2023-07-26 14:39:49