Procjena teško mjerljivih procesnih veličina na temelju pogonskih podataka

doctoral thesis

• Title (english): Estimation of difficult-to-measure process variables based on plant operational data
• Author: Dražen Slišković
• Mentor: Nedjeljko Perić (mentor)
• Committee member: Ivan Petrović (član povjerenstva)
• Committee member: Nedjeljko Perić (član povjerenstva)
• Committee member: Franjo Jović (član povjerenstva)
• Committee member: Radoslav Galić (član povjerenstva); MBZ: 81682
• Committee member: Damir Seršić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Electrical Engineering and Computing; (Department of Control and Computer Engineering); Zagreb
• Defense date and country: 2005-11-28, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Electrical Engineering; Automation and Robotics
• Universal decimal classification (UDC): 621.3 - Electrical engineering

Abstract

Primjenom estimatora moguće je nadomjestiti mjerenje teško-mjerljive procesne veličine. Estimacija se pri tome oslanja na informacije o drugim procesnim veličinama koje se mjere senzorima, tzv. lako-mjerljivim veličinama. Kako bi se estimacija mogla kontinuirano provoditi potrebno je raspolagati matematičkim modelom procesa koji povezuje odabrane lako-mjerljive veličine s teško-mjerljivom veličinom koja se estimira. Budući da u praksi najčešće nema dovoljno a priori znanja o procesu, izgradnja modela procesa se zasniva na podacima dobivenim mjerenjem. Pri tome se polazi od neke opće strukture modela koja može aproksimirati bilo koju funkciju ulazno-izlaznog preslikavanja a parametri modela se procjenjuju na temelju raspoloživih mjernih podataka.
S ekonomskog stajališta najpovoljnija je izgradnja modela procesa na temelju pogonskih podataka sadržanih u procesnoj bazi podataka. Kada se modelira na ovim podacima uobičajeni pristupi izgradnji modela često zapadaju u ozbiljne matematičke probleme i rezultiraju modelom loših predikcijskih svojstava. Za izgradnju modela na pogonskim podacima u radu se predlažu metode zasnovane na projekciji ulaznog prostora u latentni prostor i regresiji na ovim novim (latentnim) varijablama. S matematičkog stajališta ovakav model predstavlja kompoziciju dviju funkcija kojima se parametri optimiraju kroz zasebne kriterije, što u pravilu omogućava bolje strukturiranje modela procesa za određenu zadaću modeliranja, kao i pouzdaniju procjenu parametara modela. Kako su svojstva modela uglavnom određena načinom preslikavanja ulaznog prostora u latentni, u radu je težište istraživanja stavljeno upravo na ovaj aspekt izgradnje modela. Nekoliko odabranih metoda preslikavanja detaljnije su istražene kroz primjere izgradnje modela dvaju procesa.

Abstract (english)

By using an estimator, it is possible to substitute the measuring of difficult-to-measure process variable in which process the estimation relies on information about other process variables that are measured by sensors (so called easy-to-measure variables). In order to perform a continuous estimation, we need a mathematical model of the process which connects the chosen easy-to-measure variables with a difficult-to-measure variable that is being estimated. As there is usually not enough preliminary knowledge about the process, the process model building is based on the available measuring data. The basis is a general model structure which can approximate any function of input-output mapping, including presumptions of model dimension and estimation of its parameters based on the available measuring data.
In economic terms, the most suitable model process building is the one based on plant data contained in the process data base. When modeling on these data the usual approaches to model building very often face serious mathematical problems and result in a model with unsatisfactory prediction properties. To build a model using plant data, the thesis proposes methods based on input space projection into a latent space and on regression on these new (latent) variables. From a mathematical point of view, this model presents a composition of two multivariable functions, the parameters of which are optimised through separate criterion functions, which allows better model structuring for a specific modeling task and more reliable model parameters estimation. As model properties are mostly defined by way of input space mapping, the thesis focuses on this aspect of model building. Several selected methods of input mapping are investigated in two cases of model process building.

• Keywords: Estimacija teško-mjerljive procesne velièine; modeliranje procesa temeljeno na podacima; pogonski podaci; metode zasnovane na projekciji u latentni prostor; viševarijantna analiza; klaster analiza; neuronske mreže
• Keywords (english): Difficult-to-measure process variable estimation; data based process modeling; plant data; projection into latent space based methods; multivariate analysis; cluster analysis; neural networks
• Language: croatian
• URN:NBN: urn:nbn:hr:168:296760
• Study programme: Title: Doctoral study programme "Electrical Engineering and Computing"; Study programme type: university; Study level: postgraduate; Academic / professional title: doktor/doktorica znanosti, po-dručje tehničkih znanosti (doktor/doktorica znanosti, po-dručje tehničkih znanosti)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Closed access
• Created on: 2019-07-16 13:23:17