Praćenje koncentracije u procesu kristalizacije primjenom neuronskih mreža

master's thesis

• Title (english): Concentration monitoring in the crystallization process using neural networks
• Author: Veronika Žlabravec
• Mentor: Nenad Bolf (mentor)
• Committee member: Nenad Bolf (predsjednik povjerenstva)
• Committee member: Jasna Prlić Kardum (član povjerenstva)
• Committee member: Željka Ujević Andrijić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Chemical Engineering and Technology; Zagreb
• Defense date and country: 2021-09-24, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Chemical Engineering

Abstract

Strogi zahtjevi i složenost procesa stvorili su potrebu za razvojem instrumenata i postupaka kojima će se proces kristalizacije moći kontinuirano pratiti i voditi. Za kontinuirano praćenje koncentracije kao jedne od najvažnijih značajki kvalitete, u posljednje vrijeme se sve više ispituju i primjenjuju optičke sonde za kontinuirano mjerenje. S prikupljenim spektralnim podatcima moguće je razviti kalibracijski model nekom od metoda strojnog učenja. U ovom radu razvijeni su i ispitani modeli neuronskih mreža za modeliranje ovisnosti koncentracije aktivne tvari ceritiniba u tetrahidrofuranu o temperaturi i spektralnim podacima dobivenim mjerenjima u UV-Vis spektru. Nizom eksperimenata, u kojima je ispitan utjecaj predobrade podataka, strukture i broja neurona, kao i prijenosne funkcije, analizirani su modeli razvijeni neuronskim mrežama. Kalibracijski modeli služit će za kontinuirano praćenje koncentracije ceritiniba u procesu kristalizacije navedenog sustava.

Abstract (english)

The strict requirements and complexity of the process have created the need to develop instruments and procedures by which the crystallization process can be continuously monitored and controlled. Recently, optical probes have been increasingly tested and applied for continuous concentration monitoring since this is one of the most important quality features. With collected spectral data, it is possible to develop a calibration model using one of the machine learning methods. In this work, application of neural networks as a method for modeling the dependence of concentration of active substance ceritinib in tetrahydrofuran on temperature and spectral data obtained by measurements in the UV-Vis spectrum was investigated. Through a series of experiments, in which the influence of data preprocessing, number of neurons, and transfer function was examined, models developed by neural networks were analyzed. The purpose of the developed calibration models is to continuously monitor the concentration in the process of crystallization of the particular system.

• Keywords: kristalizacija; procesna analitička tehnologija; UV-Vis spektrometrija; neuronska mreža; ceritinib
• Keywords (english): crystallization; process analytical technology; UV-Vis spectrophotometry; neural network; ceritinib
• Language: croatian
• URN:NBN: urn:nbn:hr:149:553155
• Study programme: Title: Chemical 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: Closed access
• Created on: 2023-04-17 12:48:14