Opin vísindi
Opin vísindi er varðveislusafn vísindaefnis og doktorsritgerða í opnum aðgangi á vegum íslenskra háskóla og Landsbókasafns Íslands - Háskólabókasafns.
Opinn aðgangur að rannsóknaniðurstöðum er í samræmi við 10. gr. laga nr. 3/2003 um opinberan stuðning við vísindarannsóknir sem og kröfur innlendra og erlendra rannsóknasjóða. Markmiðið með opnum aðgangi er að niðurstöður rannsókna séu aðgengilegar sem flestum óhindrað og án endurgjalds á rafrænu formi. Vistun í varðveislusafninu er varanleg og ætlað að tryggja aðgang að vísindaefni íslenskra háskóla í opnum aðgangi um ókomna tíð. Varðveislusafnið Opin vísindi er tengt við rannsóknagáttina IRIS og rannsóknaniðurstöður í opnum aðgangi sem eru skráðar í IRIS eru um leið vistaðar og gerðar aðgengilegar til framtíðar í varðveislusafninu. Með því að safna þessu efni saman í eitt safn verður aðgangur að því einfaldur og þægilegur fyrir alla sem vilja kynna sér það og geta þannig notið þess öfluga vísindastarfs sem fram fer í háskólum landsins.
Varðveislusafnið er OpenAIRE / OpenAIREplus samhæft og samrýmist kröfum sem gerðar eru um birtingu rannsóknaniðurstaðna úr verkefnum sem styrkt eru úr evrópsku rannsóknaáætlununum FP7 og H2020.
Varðveislusafnið notar opna hugbúnaðinn DSpace.
Opinn aðgangur að rannsóknaniðurstöðum er í samræmi við 10. gr. laga nr. 3/2003 um opinberan stuðning við vísindarannsóknir sem og kröfur innlendra og erlendra rannsóknasjóða. Markmiðið með opnum aðgangi er að niðurstöður rannsókna séu aðgengilegar sem flestum óhindrað og án endurgjalds á rafrænu formi. Vistun í varðveislusafninu er varanleg og ætlað að tryggja aðgang að vísindaefni íslenskra háskóla í opnum aðgangi um ókomna tíð. Varðveislusafnið Opin vísindi er tengt við rannsóknagáttina IRIS og rannsóknaniðurstöður í opnum aðgangi sem eru skráðar í IRIS eru um leið vistaðar og gerðar aðgengilegar til framtíðar í varðveislusafninu. Með því að safna þessu efni saman í eitt safn verður aðgangur að því einfaldur og þægilegur fyrir alla sem vilja kynna sér það og geta þannig notið þess öfluga vísindastarfs sem fram fer í háskólum landsins.
Varðveislusafnið er OpenAIRE / OpenAIREplus samhæft og samrýmist kröfum sem gerðar eru um birtingu rannsóknaniðurstaðna úr verkefnum sem styrkt eru úr evrópsku rannsóknaáætlununum FP7 og H2020.
Varðveislusafnið notar opna hugbúnaðinn DSpace.
Flokkar í Opnum vísindum
Veldu flokk til að skoða.
- University of Iceland
- University of Akureyri
- Bifröst University
- Hólar University College
- Reykjavík University
- IRIS
- Agricultural University of Iceland
- National and University Library of Iceland
- Iceland University of the Arts
Nýlega bætt við
A Methodological Framework for the Assessment of Knowledge Risks
(2025) Foli, Samuel; Durst, Susanne; Department of Business and Economics
Knowledge risk management (KRM) has emerged as an essential field dedicated to addressing the various risks associated with organisational knowledge. A significant aspect of KRM is the assessment of knowledge risks; however, this area remains relatively underexplored due to the absence of a comprehensive framework. This PhD research aims to develop a comprehensive risk assessment framework for knowledge risks, incorporating Multi-Criteria Decision-Making (MCDM) models. Article I identifies seven key knowledge risks in ICT-supported collaborative projects using Total Interpretive Structural Modelling (TISM) as one of the MCDM models. It highlights cybercrime and espionage as high-driving risks, establishing hierarchical interrelations among these risks and providing a structural model that facilitates systematic KRM. Article II evaluates operational knowledge risks in SMEs using a grey- Decision-Making Trial and Evaluation Laboratory (DEMATEL) model. It identifies 11 critical risks, categorising them as causal (e.g., knowledge waste and gaps) and effect risks (e.g., relational risks and espionage). Outsourcing risks and improper knowledge application are identified as particularly significant. This categorisation aids SMEs in understanding and addressing their unique vulnerabilities effectively. Article III focuses on knowledge leakage drivers in collaborative agreements using an integrated ISM-MICMAC model. It identifies nine key drivers, including incomplete contracts and horizontal competition, as critical risk factors. The study demonstrates how these drivers influence knowledge leakage and provides actionable insights for mitigating risks. In the final complementary study, the research integrates TISM, DEMATEL, and Preference Ranking Organisation Method for Enrichment Evaluation (PROMETHEE) within its framework. The framework was tested using a case company to validate its practical use in assessing knowledge risks and in supporting informed decision-making. Methodologically, this thesis contributes to the literature on KRM, particularly in the domain of knowledge risk assessment, by introducing a well-structured and promising framework. •TISM is applied to capture the interdependencies among knowledge risks and knowledge risk factors. •DEMATEL is used to weight extended criteria, providing a deeper understanding of their significance. •PROMETHEE is applied to prioritise knowledge risk factors based on their level of importance. The insights derived from this framework offer valuable guidance for managers, risk managers, CEOs, and business owners, enabling them to identify, analyse and evaluate, a wide range of knowledge risks effectively. This thesis not only fills a critical gap in the KRM, more specifically knowledge risk assessment literature but also provides a practical tool to enhance decision-making and organisational resilience.
Computational methods for autonomous multirotor drone landing
(2025-06-15) Springer, Joshua David; Kyas, Marcel; Kyas, Marcel; Department of Computer Science
This dissertation presents findings on the topic of autonomous multirotor drone landing — one basic area of multirotor drone flight that is not yet fully automated — with emphasis on real world proofs of concept. We conduct two phases of research, focusing first on landing sites that are structured with fiducial markers, and next on unstructured landing sites, where the drone cannot expect to detect existing infrastructure. The first phase is a continuation of the author’s master thesis which proposes an autonomous landing method based on fiducial markers and a gimbal mounted camera that is tested in simulation. We migrate the method from simulation to the real world and then expand on it. The initial migration revealed problems in recognizing the orientation of the fiducial markers in the real world, which was obscured by idealized graphics in simulation. We quantify this orientation ambiguity in several fiducial systems and carry out a real world landing experiment without mitigating the orientation ambiguity to gauge its effects. Finally, we develop a method for avoiding this issue entirely by directing the drone based solely on the angle from the drone to the landing pad. We demonstrate this method in the real world with both visual and infrared fiducial markers. The infrared markers can serve as landing site identification infrastructure both at daytime and nighttime, and can even be unpowered. Our overall contributions on this front are 1. some modifications to existing fiducial systems to mitigate orientation ambiguity and decrease runtime computational requirements, 2. a test of the effects of orientation ambiguity on the feasibility of autonomous landing when depending primarily on fiducial pose estimation, and 3. a minimalistic method for autonomous drone landing that uses fewer data points than existing methods (primarily the pixel position of the marker and orientation of the gimbal, and avoiding use of the altitude and range) to avoid the orientation ambiguity issue and allow flexibility in changing between modalities (visible and infrared) and changing the landing pad size without reconfiguration. The second phase involves analyzing the terrain beneath the drone to determine if it is safe for landing. We focus on suburban environments as an easier test case, and on lava fields as a more challenging test case that is plentiful and relevant in Iceland. We develop a pipeline for creating appearance based terrain classifiers to automatically locate safe landing sites with a typical, monocular camera using supervised learning methods for image segmentation. To avoid hand labeling thousands of images, we generate a synthetic training data set from geometric surveys in analog environments — similar to the target landing environments, but not exactly the same. We train a U Net to do this task, but the pipeline is modular such that it is possible to use other methods. We evaluate the method synthetically and validate it on real world data to ensure that it has not overfitted to the synthetic data. The method shows success in classifying the training and validation sets in both scenarios, but can provide erroneous classifications when obstacles are too far away, or when obstacles cannot be easily identified by their appearance (e.g., safe, level gravel safe versus unsafe gravel slanted at 45 deg). We therefore supplement this method with a geometric check that uses a depth camera at low altitude (3.5 m – 5 m) which can prevent unsafe landing at erroneously classified sites. Our contributions on this front are: 1. a full pipeline to create visual terrain classifiers from the data collection stage to the real world deployment stage, 2. guidelines and scripts for automatically creating the custom synthetic data sets (particularly useful for environments with few or no publicly available topographical data sets), 3. a successful terrain classifier that goes beyond previous methods by being relatively tiny (1 2 MB), operating at varying camera angles and which is applicable to any environment for which it has training data (not limited to, e.g., urban environments), and 4. a real world demonstration of the method onboard a drone in real time. Real world testing requires a lot of engineering overhead, which itself is useful for other researchers. Therefore, we describe our systems and payloads in detail so they can be reproduced by others. We carry out autonomous landing experiments using a DJI Spark and the DJI Mobile SDK (MSDK) for autonomous control, which gives insight into how to add custom computers into the control loop when using the MSDK. Finally, we develop 3 payloads for the DJI Matrice 350 that add onboard computers with varying degrees of integration and autonomous control, and we provide guidelines for reproducing them. We target the Raspberry Pi series as the main computing board for these payloads in the hopes that the wider community of drone users can take advantage of them in a number of different fields with relative ease, given the already booming Raspberry Pi community that ensures compatibility with a huge range of other hardware and software. Our contributions on this front are: 1. guidelines for creating a fully integrated payload for the DJI Matrice series and other DJI drones, designs for the 3D printed case for such payload with quick release brackets for easy installation and removal in the field, and many field tests and demonstrations of the payload while testing the autonomous landing methods mentioned above.
High Current Arc Modeling for Silicon Submerged Arc furnaces
(2025-06-12) Haraldsson, Hákon Valur; Sævarsdóttir, Guðrún Arnbjörg; Tesfahunegn, Yonatan Afework; Tangstad, Merete; Reynolds, Quinn; Department of Engineering
Electric arcs are a necessary heat source in many industrial processes that take place in Submerged Arc Furnaces (SAFs). Arcs exhibit non-linear electrical characteristics and behave in a complex manner. Therefore, an improved understanding of their behavior enables better control of furnace operation. Modeling of industrial arcs is a multiphysics process that involves simultaneously solving several coupled physical phenomena, such as electromagnetics, uid dynamics and heat transfer, including a radiative heat transfer from the plasma arc. Coupling uid dynamics and electromagnetics is known as Magnetohydrodynamics (MHD). However, there are also simpler approaches to arc modeling, either based on simplifed physical principles or empirical behaviour. Direct measurement of the arc characteristics is impossible due to hostile conditions inside the SAF, so controlling the heat dissipation is both a science and an art. The arcs exhibit non-linear electrical characteristics and behave in a complex manner. We start by discussing a DAQ system gathering data from a FeSi SAF, the data is processed and used to determine various furnace conditions including arc and charge current as well as harmonics. In this work, several computational models for arc are implemented. First a combined Cassie-Mayr model (CMM) and a channel arc model (CAM), are implemented and coupled with a submerged arc furnace electrical circuit model. The complete circuit model parameters such as resistances and inductances are estimated using measurements conducted on an operational furnace which are also used to validate the models. Both models are then used to estimate harmonic distortion in a SAF for different arc power ratios. Secondly a MHD model implemented by the author is used to simulate alternating current arcs with different plasma gas compositions, and compared to second MHD model. The thermophysical properties of each composition are calculated using specialized code as well as gathered from literature. We investigate the dependence of the results on both the MHD models used and the input plasma data for three argon data sets and compare the results to data obtained from laboratory experiments. Finally, we investigate furnace conditions using different ratios of SiO to CO gases. Finally a new implementation of a special sub model for cathode and anode surface conditions is presented. The models is first used with the channel arc model and then integrated into the MHD model as a module. This model is used to investigate the electrode erosion as well as the sheat voltage present close the the plasma wall.
Nickel–Iron–Copper-Based Oxygen Evolving Anode for Low Temperature Aluminum Electrolysis
(2025-04-29) Singh, Kamaljeet; Sævarsdóttir, Guðrún Arnbjörg; Gunnarsson, Gudmundur; Haarberg, Geir Martin; Department of Engineering
The Hall–Héroult process is, currently, the only industrial method for primary aluminum production. However, the process suffers from many inefficiencies, mainly because the carbon anodes are continuously consumed during electrolysis, producing greenhouse gas emissions—primarily CO2 and, intermittently, perfluorocarbons. Another inefficiency in the process is its high energy demand, essentially caused by a significant ohmic voltage drop resulting from the large anode-cathode distance created by the molten aluminum pool, and a high anodic overpotential due to the slow kinetics of the anodic process. Therefore, to eliminate greenhouse gas emissions and improve energy utilization in aluminum electrolysis a non-consumable, cost effective, and efficient oxygen evolving anode (OEA) is essential. Recent research has shown that the use of nickel–iron–copper-based alloys for the OEA offers promising performance in low temperature electrolytes. This performance advantage is attributed to their ability to form a protective nickel-ferrite (NiFe2O4) oxide scale during anodic reactions, coupled with their reduced wear rates under these operating conditions. Unfortunately, systematic studies on the effects of alloy composition and low temperature electrolyte composition on the formation and stability of the protective scale are lacking. The present work explored the use of various earth-abundant Ni–Fe–Cu-based alloys for the OEA in a range of low temperature KF-NaF-AlF3-Al2O3(sat.)-based electrolyte compositions for aluminum electrolysis at 800 ⁰C. Additionally, a TiB2 wettable aluminum cathode and a vertical electrode configuration were employed to develop a compact and energy efficient cell. To identify the optimal compositions and conditions for aluminum electrolysis in a 40 A laboratory cell, two electrolyte compositions, K-rich and Na-rich, were investigated using Ni–Fe–Cu alloys as anodes. It was found that the K-rich electrolyte composition in combination with Ni42-Fe38-Cu20 anode offered a low anode wear rate and stable electrolysis. This performance was attributed to the better alumina solubility of the electrolyte and the formation of a dense and protective NiFe2O4 oxide on the anode surface. The oxidation treatment of the Ni42-Fe38-Cu20 alloy, to pre-form an oxide scale, demonstrated its ability to form a multi-layered oxide scale of CuO, Fe2O3 and protective NiFe2O4. This indicated the effectiveness of the treatment in developing a protective oxide scale ex-situ, which was found satisfactory for meeting the requirements of the OEA. The use of the OEA leads to higher energy demands compared to the Hall–Héroult process with carbon anodes, primarily due to the increased reaction voltage. Therefore, to assess the energy efficiency in terms of overpotential on OEA, steady state anodic polarization curves were obtained on platinum and a series of Ni–Fe–Cu-based alloys. The polarization curve on the platinum anode exhibited two linear regions, showing good consistency with the proposed theoretical mechanism of oxygen evolution reaction. The polarization curve on alloys, both in oxidized and untreated conditions, however, exhibited a single Tafel region. At a normal current density of 0.8 Acm-2, the oxidized anodes Ni42-Fe38-Cu20 and Ni48-Fe47-Cu5 showed lower overpotentials after electrolysis compared to untreated anodes of same composition, respectively. This resulted from the fact that oxidized anodes exhibited better electrocatalytic activity with lower Tafel slopes, mainly due to the pre-formed conductive oxide scale through oxidation treatment.
Language Representation Models for Low- and Medium-Resource Languages
(2025) Daðason, Jón Friðrik; Loftsson, Hrafn; Department of Computer Science
Transformer-based language models have proven to be extremely effective for a wide variety of natural language understanding tasks, including question answering, automatic text summarization, and sentiment analysis. These models are typically pre-trained on large, unannotated corpora using self-supervised tasks such as masked token prediction, often requiring weeks or months of training, followed by fine-tuning on practical tasks, which requires substantially less time and data by comparison. Since their introduction, Transformer models have grown exponentially in size, from approximately 100 million parameters in 2018 to over 600 billion in 2024, with the largest pre-training corpora growing from around 800 million tokens to over 14.8 trillion. However, many low- and medium-resource languages lack the extensive datasets and computational resources required to pre-train language models at this scale. Therefore, data-efficient pre-training techniques are crucial for effectively utilizing the limited resources available for these languages. In this thesis, we investigate various data-efficient pre-training strategies and evaluate their impact on downstream tasks in six low- to medium-resource languages: Icelandic, Estonian, Basque, Galician, Nepali, and Tajik. First, we analyze several text quality filtering techniques to discard noisy data from web-crawled corpora. We propose a novel, language-independent filtering approach using unsupervised clustering and outlier detection algorithms which achieves comparable performance to a rule-based approach. Second, we explore the effects of augmenting monolingual pre-training corpora with text from related and unrelated languages, as well as Python code, finding significant improvements in downstream performance for certain tasks for larger models. Our results support the hypothesis that linguistic similarity facilitates cross-lingual transfer. Finally, we compare several subword tokenization algorithms and evaluate their impact on downstream results when used in pre-trained language models. Our analysis reveals that the Unigram algorithm consistently yields the best results on downstream tasks, and that a vocabulary size of 64k outperforms smaller vocabularies by a statistically significant margin. Our findings demonstrate that data-efficient pre-training techniques can substantially improve the performance of language models for low- and medium-resource languages. By optimizing the use of available data and resources, we achieve statistically significant improvements in downstream tasks under data-constrained conditions, paving the way for more effective natural language processing in resource-constrained settings. We release several datasets and tools compiled and developed during the work of this thesis, as well as multiple pre-trained Transformer-based language models.