Željka Ujević Andrijić

Mrs. Željka Ujević Andrijić

orcid

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
9 August 2025 in Sensors
Verified
Journal Article
Open access
PDF file
ŽAŽeljka Ujević Andrijić
Željka Ujević Andrijić
NRNikola Rimac
Nikola Rimac
Fouling detection in refinery crude distillation unit (CDU) preheat trains is essential for maintaining energy efficiency and operational reliability. This study presents a virtual sensing approach for fouling monitoring using data-driven and semi-empirical models. Specifically, Long Short-Term Memory (LSTM) neural networks, Extreme Gradient Boosting (XGB), and the ɛ-NTU method (effectiveness—Number of Transfer Units) were compared for predicting heat exchanger outlet temperatures, which serve as indicators of fouling. Models were trained on clean operational data to estimate baseline performance. A growing discrepancy between predicted and actual outlet temperatures over time indicated heat transfer degradation. Fouling resistance was calculated from the difference between predicted and actual heat transfer coefficients, enabling effectiveness loss assessment. The LSTM model showed high accuracy in capturing dynamic operational trends, while XGB provided a lightweight alternative with limited extrapolation capability under unfamiliar conditions. Both models outperformed the ɛ-NTU approach in fouling detection sensitivity. Inefficiencies from a single fouled exchanger were estimated to result in an additional 175 tons of CO2 emissions and an economic loss of approximately EUR 12,000 over two months. This study highlights the potential of AI-enabled virtual sensors for real-time fouling monitoring in industrial heat exchangers. Such tools can significantly enhance predictive maintenance strategies, improve energy efficiency, and reduce emissions.

16 December 2023 in Processes
Verified
Journal Article
Open access
PDF file
MGMatea Gavran
Matea Gavran
ŽAŽeljka Ujević Andrijić
Željka Ujević Andrijić
NBNenad Bolf
Nenad Bolf
NRNikola Rimac
Nikola Rimac
JSJosip Budimir Sacher
Josip Budimir Sacher
DŠDamir Šahnić
Damir Šahnić
Raman spectroscopy is a useful tool for polymorphic form-monitoring during the crystallization process. However, its application to solute concentration estimation in two-phase systems like crystallization is rare, as the Raman signal is influenced by various changing factors in the crystallization process. The development of a robust calibration model that covers all variations is complex and represents a major challenge for the implementation of Raman spectroscopy for in-line monitoring and control of the solution crystallization process. This paper describes the development of a Raman-based calibration model for estimating the solute concentration of the active pharmaceutical ingredient ceritinib. Several different calibration approaches were tested, which included both temperature and spectra of clear solutions and slurries/suspensions. It was found that the concentration of the ceritinib solution could not be accurately predicted when suspended crystals were present. To overcome this challenge, the approach was enhanced by including additional variables related to crystal size and solid concentration obtained via in-line process microscopy (chord-length distribution percentiles D10, D50 and D90) and turbidity. Partial least squares regression (PLSR) and artificial neural network (ANN) models were developed and compared based on root mean square error (RMSE). ANN models estimated the solute concentration with high accuracy, with the prediction error not exceeding 1% of the nominal solute concentration.

2 November 2023 in Mathematics
Verified
Journal Article
Open access
PDF file
SHSrečko Herceg
Srečko Herceg
ŽAŽeljka Ujević Andrijić
Željka Ujević Andrijić
NRNikola Rimac
Nikola Rimac
NBNenad Bolf
Nenad Bolf
Dynamic neural networks (DNNs) are a type of artificial neural network (ANN) designed to work with sequential data where context in time is important. Unlike traditional static neural networks that process data in a fixed order, dynamic neural networks use information about past inputs, which is important if the dynamic of a certain process is emphasized. They are commonly used in natural language processing, speech recognition, and time series prediction. In industrial processes, their use is interesting for the prediction of difficult-to-measure process variables. In an industrial isomerization process, it is crucial to measure the quality attributes that affect the octane number of gasoline. Process analyzers commonly used for this purpose are expensive and subject to failure. Therefore, to achieve continuous production in the event of a malfunction, mathematical models for estimating product quality attributes are imposed as a solution. In this paper, mathematical models were developed using dynamic recurrent neural networks (RNNs), i.e., their subtype of a long short-term memory (LSTM) architecture. The results of the developed models were compared with the results of several types of other data-driven models developed for an isomerization process, such as multilayer perceptron (MLP) artificial neural networks, support vector machines (SVM), and dynamic polynomial models. The obtained results are satisfactory, suggesting a good possibility of application.

19 August 2025 in Actuators
Verified
Journal Article
Open access
PDF file
ŽAŽeljka Ujević Andrijić
Željka Ujević Andrijić
SHSrečko Herceg
Srečko Herceg
MŠMagdalena Šimić
Magdalena Šimić
NBNenad Bolf
Nenad Bolf
This work presents the development of soft sensor models for monitoring the operation of online process analyzers used to measure the sulfur content in the product of the refinery hydrodesulfurization process. Since sulfur content often fluctuates over time, soft sensor models must account for these frequency fluctuations. We have therefore developed dynamic data-driven models based on linear and nonlinear system identification techniques (finite impulse response—FIR, autoregressive with exogenous inputs—ARX, output error—OE, nonlinear ARX—NARX, Hammerstein–Wiener—HW) and machine learning techniques, including models based on long short-term memory (LSTM) and gated recurrent unit (GRU) networks, as well as artificial neural networks (ANNs). The core steps in model development included the selection and preprocessing of continuously measured plant process data, collected from a full-scale industrial hydrodesulfurization unit under normal operating conditions. The developed soft sensor models are intended to support or replace process analyzers during maintenance periods or equipment failures. Moreover, these models enable the application of inferential control strategies, where unmeasured process variables—such as sulfur content—can be estimated in real time and used as feedback for advanced process control.

19 November 2023 in Kemija u industriji
Verified
Journal Article
Open access
THTesa Herceg
Tesa Herceg
ŽAŽeljka Ujević Andrijić
Željka Ujević Andrijić
MGMatea Gavran
Matea Gavran
JSJosip Budimir Sacher
Josip Budimir Sacher
IVIvan Vrban
Ivan Vrban
NBNenad Bolf
Nenad Bolf
Procesna analitička tehnologija (PAT) sve se češće primjenjuje u procesu kristalizacije za kontinuirano praćenje neke od ključnih procesnih veličina i značajki kvalitete proizvoda. Vrlo bitne procesne varijable za kristalizaciju hlađenjem su temperatura i koncentracija otopine. Kontinuirano mjerenje koncentracije omogućeno je naprednim in situ spektroskopskim instrumentima. U takve metode spada i prigušena ukupna reflektancija Fourier transformacijske infracrvene spektroskopije (engl. Attenuated Total Reflectance Fourier Transform Infrared Spectroscopy, ATR-FTIR), primijenjena u ovom istraživanju. U ovom radu razvijeni su kemometrijski modeli za kontinuiranu procjenu koncentracije otopine u stvarnom vremenu metodama strojnog učenja. Cilj ovog rada bio je razviti i ispitati modele parcijalne regresije metodom najmanjih kvadrata (engl. Partial Least Squares Regression, PLSR) i modele neuronskih mreža (engl. Neural Networks, NN) za modeliranje ovisnosti koncentracije djelatne tvari, ksilometazolin hidroklorid, u n-butanolu, o temperaturi i spektralnim podatcima dobivenim mjerenjima ATR-FTIR spektrometrom. U radu je provedena predobrada prikupljenih podataka tehnikom MSC (engl. Multiplicative Scatter Correction) te je provedeno skaliranje podataka prema normalizacijama Min-Max i Z-score. Analiziran je broj neurona u prvom i drugom skrivenom sloju, broj skrivenih slojeva, vrsta primijenjenog algoritma učenja (ADAM, NADAM, RMSprop) kao i utjecaj vrste prijenosne funkcije (ReLU, sigmoid, tanh) na kvalitetu razvijenih neuronskih mreža. Razvijeni modeli na sva četiri skupa podataka dali su iznimno dobre rezultate s obzirom na iznose koeficijenta determinacije i srednje kvadratne pogreške. Modelom neuronske mreže dobiveni su koeficijenti determinacije u rasponu vrijednosti od 0,9979 do 0,9989, dok je srednja kvadratna pogreška od 0,0020 do 0,0011. PLSR modelom dobiveni su koeficijenti determinacije od 0,9990 do 0,9995, odnosno srednja kvadratna pogreška od 0,0009 do 0,0005. Dobiveni rezultati pokazali su da predobrada podataka i dodavanje drugoga skrivenog sloja neuronske mreže u ovoj studiji nisu imali velik utjecaj na konačne rezultate. Takav način praćenja i kontrole procesa rezultira efikasnijom proizvodnjom uz smanjenu vjerojatnost pogreške, što u konačnici farmaceutskoj industriji omogućuje brže plasiranje proizvoda na tržište.

1 October 2023 in Preprints.org
Verified
Preprint
Open access
PDF file
SHSrečko Herceg
Srečko Herceg
ŽAŽeljka Ujević Andrijić
Željka Ujević Andrijić
NRNikola Rimac
Nikola Rimac
NBNenad Bolf
Nenad Bolf
Dynamic neural networks (DNN) are types of artificial neural networks (ANN) that are designed to work with sequential data where context in time is important. In contrast to traditional static neural networks that process data in a fixed order, dynamic neural networks use information about past inputs, which is important if dynamic of a certain process is emphasized. They are widely used in natural language processing, speech recognition and time series prediction. In industrial processes, their use is interesting for the prediction of difficult-to-measure process variables. In an industrial process of isomerization, it is crucial to measure the quality attributes affecting the octane number of gasoline. Process analyzers that are commonly used for this purpose are expensive and subject to failures, therefore, in order to achieve continuous production in case of malfunction, mathematical models for estimating the product quality attributes are imposed as a solution. In this paper, mathematical models were developed using dynamic recurrent neural networks (RNN), i.e., their subtype of a long short-term memory (LSTM) architecture. The results of the developed models were compared with the results of several types of other data-driven models developed for an isomerization process, such as multilayer perceptron (MLP) artificial neural networks, support vector machines (SVM) and dynamic polynomial models. The obtained results are satisfactory, which suggests a good possibility of application.