NTNU Open (Norwegian University of Science and Technology)
Not a member yet
92030 research outputs found
Sort by
Verdsettelse av Borregaard ASA
No full textOppgavens mål og hensikt er å finne ut hvor mye Borregaard ASA er verdt den 24.03.2025. For å besvare oppgaven har det blitt gjennomført ulike strategiske analyserer, som fokuserer på både eksterne og interne faktorer. De eksterne faktorene ble analysert gjennom Porters bransjeanalyse (fem krefter) og en PESTEL-analyse. Hensikten med de strategiske analysene er for å vurdere konkurransebildet til Borregaard, finne ut mer om businessmodellen og om selskapet har «moat» i forhold til deres konkurrenter. Videre har vi beregnet et estimat for selskapsverdien til Borregaard basert på de frie kontantstrømmene til totalkapitalen, samt gjennom en multippelbasert verdsettelse.
Oppgaven har benyttet seg av historiske regnskapstall fra 2019 og frem til 2024 når man både har analysert nøkkeltall, samt gjennom estimering av de kommende kontantstrømmene i årene fremover. Gjennom regnskapsanalysen er det spesielt EBITDA-margin forbedring som er bemerkelsesverdig. Under Covid-19 fikk Borregaard kjenne på økte materialkostnader, tømmerkostnader og svært volatile energipriser. Til tross for dette forbedret EBITDA-marginen seg betraktelig og dette er spesielt på grunn av en imponerende evne til å flytte kostnader over på kunden. Det er også verdt å bemerke seg at avkastningen på sysselsattkapital er over målet til selskapet selv på 15%.
Beskrivelsen av Borregaards evne til å overføre kostnader til sluttkunden, sammen med spesialiseringsstrategien, er lagt til grunn for regnskapsestimatene fremover i tid, der vi antar en margin-ekspansjon. Det er lagt fokus på perioden 2025 til 2027 ettersom Borregaard har estimater for kapitalkostnader i denne perioden. Avkastningskravet på totalkapitalen (WACC) er beregnet til 10,01% og vi har brukt en vekstfaktor inn i evigheten på 2%.
Avslutningsvis har vi vektlagt den frie kontantstrømanalysen og exit-multippel verdsettelsen 20/80 og vi har kommet frem til følgende kursmål på 208 NOK for Borregaard. aksjekursen den 24.03.2025 ligger på 170 som impliserer en oppside på 20%. Vedlagt ligger også ulike sensitivitetsanalyser for både evig vekstfaktor, WACC, beta og exit-multippel. Vi mener det ligger oppside i Borregaard ASA, basert på estimater, historikk og kvalitet i forhold til konkurrerende selskaper.The objective of this paper is to determine the value of Borregaard ASA as of March 24, 2025. To address this, various strategic analyses have been conducted, focusing on both external and internal factors. The external factors were analysed using Porter’s industry analysis (Five Forces) and a PESTEL analysis. The purpose of these strategic analyses is to assess Borregaard’s competitive landscape, understand its business model, and determine whether the company has a “moat” relative to its competitors. Furthermore, we have estimated Borregaard’s company value based on free cash flows to total capital, as well as through a multiple-based valuation.
The analysis utilizes historical financial statements from 2019 to 2024 to examine key financial ratios and estimate future cash flows. The financial statement analysis highlights a particularly noteworthy improvement in the EBITDA-margin. During COVID-19, Borregaard experienced increased material costs, timber costs, and highly volatile energy prices. Despite these challenges, the EBITDA-margin improved, primarily due to the company’s strong ability to pass costs onto customers. Additionally, it is worth noting that the return on capital employed exceeds the company’s own target of 15%.
Borregaard’s ability to transfer costs to end customers together with the specialization strategy has been a key assumption in our financial estimates, where we assume margin expansion going forward. The analysis focuses on the period from 2025 to 2027, as Borregaard has provided capital expenditure estimates for this timeframe. The required rate of return on total capital (WACC) has been calculated at 10.01%, and a perpetual growth rate of 2% has been applied.
In conclusion, we have weighted the free cash flow analysis and the exit multiple valuation at 20/80, arriving at a target price of 208 NOK for Borregaard. The stock price as of March 24, 2025, is 170 NOK, implying an upside potential of 20%. Additionally, various sensitivity analyses have been conducted for the perpetual growth factor, WACC, beta, and exit multiple. Based on estimates, historical performance, and quality relative to competitors, we believe there is upside potential in Borregaard ASA
Disrupsjon på legevakten Kritiske team- og ledelsesfaktorer for å lykkes med en disrupsjonsstrategi i et programvare-startupselskap innen helsesektoren.
No full textSammendrag
Norske myndigheters strategi for realisering av «én innbygger, én journal» har gått fra en offentligdrevet tilnærming til en mer åpen modell, der private aktører kan spesialisere seg og konkurrere i markedet. Teknologiskift skaper muligheter for startup-selskaper til å utfordre etablerte aktører gjennom disrupsjon.
Studien er en enkeltcasestudie av selskapet Kvikna, som utvikler et elektronisk pasientjournalsystem for legevakt. Datainnsamlingen bygger på 17 semistrukturerte intervjuer med ledere i Kvikna og tilknyttede selskaper og er analysert med tematisk analyse (Braun & Clarke, 2006). Problemstillingen for oppgaven er: Hva er kritiske team- og ledelsesfaktorer for å lykkes med en disrupsjonsstrategi i et programvare-startupselskap innen helsesektoren?
Vi har utviklet begrepet kritiske team- og ledelsesfaktorer (KTL-faktorer). Vi har identifisert tre KTL-faktorer: (i) team med balanse mellom teamroller, (ii) team med tilstrekkelig formålsnivå, og (iii) organisering som sikrer tilstrekkelig agilitet. KTL-faktorene må understøtte evnen til løse de kritiske suksessfaktorene (KSF).
Drøftingen viser at KTL-faktorene henger tett sammen med KSFene, og at disrupsjon i helsesektoren krever deltakelse av alle teknologi- og teamroller, samt at teamene må virke på et høyt formålsnivå for å lykkes med en disrupsjonsstrategi i et programvare-oppstartsselskap i helsesektoren.
Drøftingen viser vi at dersom et team ikke kan lykkes med å gjennomføre strategien, så vil en finne forklaringer i sammensetningen av team, formålsnivået til teamet evner og i hvilken grad agil organisering praktiseres på en måte som sikrer effektive avklaringer i hele organisasjonen.
Studien er et bidrag til praktikere og forskning på sammenhengen mellom strategi og team- og ledelsesfaktorer for å lykkes med en disrupsjonsstrategi for oppstartsselskaper som vil levere til helsetjenesten.
Den åpenbare svakheten ved vår drøfting og våre resultater, er at empirien vår kun er basert på ett enkeltcase gjennomført på et svært tidlig stadium i det studerte selskapets livsløp. På tross av svakheter så kan studien likevel være et utgangspunkt for videre forskning. For eksempel kan studien gjentas om noe tid og/eller i andre selskap, eller som fler-case-studie for å sammenligne resultater. Et annet tema for videre forskning kan være om KTL-faktorer kan også kan anvendelses på flere typer strategier eller sektorer.
Søkeord: Team, Technology Strategy, Entrepreneurial Strategy, Agile, Startup, E-HealthAbstract
The strategy of Norwegian authorities for realizing the vision of "one citizen, one record" has transitioned from a publicly driven approach to a more open model, where private actors can specialize and compete in the market. Technological shifts create opportunities for startup companies to challenge established players through disruption.
This study is a single case study of the company Kvikna, which develops an electronic patient record system for emergency care. Data collection is based on 17 semi-structured interviews with managers at Kvikna and affiliated companies, analyzed using thematic analysis (Braun & Clarke, 2006). The research question addressed in this study is: What are the critical team and leadership factors for successfully implementing a disruption strategy in a software startup company within the healthcare sector?
We have developed the concept of critical team and leadership factors (CTL factors). We have identified three CTL factors: (i) teams with a balance of roles, (ii) teams with a sufficient level of purpose, and (iii) organizational structures that ensure adequate agility. These CTL factors must support the ability to address critical success factors (CSFs).
The discussion demonstrates that CTL factors are closely linked to CSFs, and that disruption in the healthcare sector requires the participation of all technology and team roles. Additionally, teams must operate at a high level of purpose to successfully implement a disruption strategy in a software startup within the healthcare sector.
The discussion further reveals that if a team fails to execute the strategy successfully, explanations can be found in the team's composition, the team's level of purpose, and the extent to which agile practices are implemented to ensure effective decision-making across the organization.
This study contributes to both practitioners and researchers by exploring the relationship between strategy and team and leadership factors in successfully implementing a disruption strategy for startups aiming to deliver solutions to the healthcare sector.
A notable limitation of our discussion and findings is that our empirical data is based on a single case study conducted at a very early stage in the lifecycle of the studied company. Despite this limitation, the study can serve as a foundation for further research. For example, the study could be repeated after some time and/or in other companies, or as a multiple case study to compare results. Another potential area for further research is whether CTL factors can be applied to other types of strategies or sectors.
Keywords: Team, Technology Strategy, Entrepreneurial Strategy, Agile, Startup, E-Healt
Impact of snow and building management on ground surface temperatures in permafrost environments - A case study from the historical mining town Ny-Ålesund, Svalbard
No full textPermafrost is warming due to changing climatic conditions, a trend that might threaten infrastructure and livelihood across the Arctic. Historical structures are especially vulnerable, as they were not designed to withstand these rapid environmental changes. In the high-Arctic settlement Ny-Ålesund, Svalbard, a large number of historical buildings exist, making the village a suitable case study for investigating the interplay between historical buildings and permafrost. This study analysed two years of ground surface temperature (GST) measurements and snow observations to assess the impact of snow and building management on permafrost conditions in Ny-Ålesund. The mean annual ground surface temperature (MAGST) was found between −1.9°C and 1.9°C in 2022/23, and between −3.1°C and 1.1°C in 2023/24. The results reveal that GST varied substantially within the village area and exceeded the freezing point in several locations. We identified influencing factors contributing to these differences in GST: (i) Snow redistribution by wind and snow ploughing caused large variations in GST with the highest MAGST beneath artificial snow deposits. (ii) Building type, particularly crawl space ventilation, also affected GST. Ventilated crawl spaces tended to lower the MAGST beneath the buildings, while enclosed crawl spaces increased MAGST to positive values. (iii) Building wall orientation further influenced GST, with southern exposures exhibiting higher values compared to northern exposures. Our findings highlight the importance of snow and building management on GST and permafrost stability. Understanding these small scale effects is crucial for the preservation of historical infrastructure on permafrost under changing climatic conditions.publishedVersio
Lærerens kunnskap i matematiske samtaler
No full textDenne masteroppgaven har undersøkt problemstillingen:
Hva slags matematikklærerkunnskap bruker lærere til å lede matematiske samtaler?
For å svare på problemstillingen har jeg undersøkt forskningsspørsmålene:
•Hva slags matematikklærerkunnskap kommer til syne når en lærer leder en matematisk samtale?
•Hva slags matematikklærerkunnskap kommer til syne i et videostimulert intervju med læreren?
Studien undersøker én lærers ledelse av matematiske samtaler på ungdomstrinnet. Studien bruker videostimulert intervju for å få tilgang på lærerens refleksjoner og begrunnelser. Datamaterialet består av 35 minutter videoopptak fra to klassesamtaler og to timer videoopptak av intervju med læreren. Oppgaven som brukes i studien er Rammeproblemet (Boaler & Humphreys, 2005), med algebra og problemløsning som tema.
Et rammeverk samensatt av teori om lærerhandlinger (Fraivillig et al., 1999), dialoghandlinger (Alrø & Skovsmose, 2006), samtaletrekk (Chapin et al., 2013) og prinsipper for samtalen (Kazemi & Hintz, 2021) ble brukt for å analysere lærerhandlingene i samtalen. Advancing Childrens Thinking-rammeverket (Fraivillig et al., 1999) sine kategorier; fremkalle, støtte og utvide brukes som overordnet struktur for analysen. Et rammeverk for matematikklærerkunnskap basert på Ball et al. (2008) og Kunnskapskvartetten (Rowland, 2013) ble brukt for å undersøke kunnskapen som ble identifisert i samtalene og de videostimulerte intervjuene.
Sentrale funn i studeien er at læreres grunnleggende holdninger til faget, læring og matematisk forståelse påvirker samtalen, elevenes bidrag og klasseromsnormer. Som en følge må utviklingsarbeidet i skoler ta hensyn til dette. Det kommer også frem at læreren bruker en kompleks kunnskap for å lede samtalene, der metodiske valg blant annet er begrunnet i kunnskap om fag og elever.
Studien bekrefter at videostimulert intervju er en metode som kan bidra til refleksjon over egen praksis og foreslås som et nyttig verktøy i utviklingsarbeid.This master’s thesis responds to the following research question:
What kind of mathematical knowledge for teaching do teachers need to lead mathematical discussions?
To answer this question, the following research questions have been examined:
•What kind of mathematical knowledge for teaching is evident during a mathematical discussion?
•What kind of mathematical knowledge for teaching is evident in a video-stimulated interview with the teacher?
The study investigates a teacher's orchestrating of mathematical discussions. The study uses video-stimulated interviews to access the teacher's reflections and justifications. The empirical data of this study consists of 35 minutes worth of video recordings of classroom discussions, as well as two hours of video recordings of interviews with the teacher The task used in the study is the Frame Problem (Boaler & Humphreys, 2005), with algebra and problem-solving as the theme.
A framework composed of theories on teacher actions (Fraivillig et al., 1999), dialogue actions (Alrø & Skovsmose, 2006), talk moves (Chapin et al., 2013), and principles of conversation (Kazemi & Hintz, 2021) was used to analyze the teacher's actions in the discussion. The Advancing Children's Thinking framework (Fraivillig et al., 1999) categories; eliciting, supporting, and extending are used as the overarching structure for the analysis. A framework for mathematical knowledge for teaching based on the theories of Ball et al. (2008) and The Knowledge Quartet (Rowland, 2013) was used to examine the knowledge identified in the discussions and the video-stimulated interviews.
Key findings of the study are that teachers' fundamental attitudes towards the subject, learning, and mathematical understanding influence the discussion, students' contributions, and classroom norms. Consequently, development work in schools must take this into account.
The teacher uses complex knowledge to lead the discussions, where methodological choices are justified by knowledge of the subject and students among other things.
The study confirms that video-stimulated interviews can contribute to reflection on one's own practice and are proposed as a useful tool in development work
The Riemannnian LTMADS to solve the Spectral Procrustes problem in Manopt.jl
No full textProkrustes problemet, som søker en optimal ortogonal transformasjon mellom to matriser, har ulike
anvendelser innen matematikk. Forskning har i stor grad fokusert på det ortogonale Prokrustes-
problemet, der målfunksjonen er definert ved Frobenius-normen. Dette problemet er godt forst˚att
og har en kjent lukket-form løsning.
I denne oppgaven undersøker vi det Spektrale Prokrustes-problemet, en mindre studert variant
der målfunksjonen er definert ved spektralnormen. Spektralnormen kan gi et annet perspektiv p˚a
Prokrustes-problemet. Denne alternative formuleringen kan i visse tilfeller avdekke situasjoner der
løsningen for det Spektrale Prokrustes-problemet er å foretrekke fremfor løsningen for det ortogo-
nale Prokrustes-problemet. Imidlertid fører Spektralnormens ikke-glatte natur til at gradientbaserte
optimaliseringsmetoder ikke er anvendelige. Videre har det Spektrale Prokrustes-problemet, i mot-
setning til Frobenius-normtilfellet, ingen kjent lukket-form løsning som eksisterer.
For å møte disse utfordringene benytter vi en gradientfri optimaliseringsmetode kjent som Lower
Triangular Mesh Adaptive Direct Search-algoritmen. Denne metoden opererer på en spesifikk Rie-
mannsk manifold, rotasjonsmanifolden, og sikrer at ortogonale begrensninger blir bevart gjennom
hele optimaliseringsprosessen.The Procrustes problem, which seeks an optimal orthogonal transformation between two matrices,
has various applications in mathematics. Research has predominantly focused on the orthogonal
Procrustes problem, where the objective function is defined by the Frobenius norm. This problem
is well-understood and has a known closed-form solution.
In this thesis, we investigate the Spectral Procrustes problem, a less-studied variant where the ob-
jective function is defined by the Spectral norm. The Spectral norm may offer a distinct perspective
on the Procrustes problem. This alternative formulation may, in some cases, reveal scenarios where
the Spectral Procrustes solution is preferred over the orthogonal Procrustes solution. However,
the Spectral norm’s non-smooth nature makes gradient-based optimization methods inapplicable.
Furthermore, unlike in the Frobenius norm case, the Spectral Procrustes problem may not have a
known closed-form solution.
To address these challenges, we employ a gradient-free optimization method known as the Lower
Triangular Mesh Adaptive Direct Search algorithm. This method operates on a specific Riemannian
manifold, the rotation manifold, ensuring that orthogonal constraints are preserved throughout the
optimization process
SnowPole Detection: A comprehensive dataset for detection and localization using LiDAR imaging in Nordic winter conditions
No full textThe SnowPole Detection dataset is a comprehensive collection of labeled LiDAR images, specifically designed for snow pole detection in road environments. This dataset was collected using a high-resolution OS2-128 LiDAR sensor mounted on an autonomous vehicle research platform, covering diverse environments such as mountainous, open, and forested areas. The SnowPole Detection dataset supports applications in computer vision, with a particular focus on snow pole detection and localization. The OS2-128 LiDAR sensor captures point clouds, which are processed using the Ouster SDK to generate 360-degree images in four modalities: Near-IR, Signal, Reflectivity, and Range. To enhance usability, color images were generated by assigning the first three modalities (Near-IR, Signal, and Reflectivity) to the blue, green, and red channels, respectively, excluding the Range modality. Initial labeling was conducted using Roboflow, with further refinement in CVAT, resulting in high-quality annotations. The dataset comprises a total of 1,954 manually labeled images, divided into 1,367 training images, 390 validation images, and 197 test images, following a 70/20/10 split. Since the images across all modalities are pixel-aligned, the labels for the color images are also applicable to each modality individually. This structure allows researchers to directly use the dataset for snow pole detection tasks, whether focusing on color or individual LiDAR modalities. The SnowPole Detection dataset is publicly available at Mendeley.publishedVersio
Local area cooling versus broad area cooling for boil-off reduction in large-scale liquid hydrogen storage tanks
Full text linkFuture use of liquid hydrogen (Image 1 ) as an effective energy carrier will require elimination or minimization of hydrogen boil-off that is not utilized by demands in the value chain. The present work promotes local area cooling (LAC) as a promising boil-off reduction technology. In contrast to the more conventional broad area cooling (BAC), LAC targets local, concentrated heat flows e.g. through tank support structures. This yields important practical benefits, especially for large-scale tanks, due to the order-of-magnitude reduction in the size of the cooling system. Such benefits include lower capital costs and simpler installation, maintenance and coolant management. LAC applied outside the outer tank wall is particularly attractive for tanks with evacuated insulation. In a series of numerical studies, we use the finite element method to evaluate the thermal performance of LAC and BAC in the context of ship-borne Image 1 transport. The studies concern 40 000 m3-capacity, skirt-supported tanks insulated using evacuated perlite or helium-filled polyurethane (HePUR) foam. For the perlite-insulated tank, LAC and BAC with liquid nitrogen coolant can reduce the daily boil-off rate from 0.04%/day to, respectively, 0.011%/day and 0.004%/day. The corresponding numbers for CO2-based refrigeration are 0.031%/day and 0.028%/day. For the HePUR-insulated tank, which has a higher baseline boil-off rate of 0.24%/day, reduced boil-off rates down to 0.17%/day and 0.04%/day are achievable using LAC and BAC, respectively. LAC and BAC both offer increased power efficiency in comparison to reliquefaction only.Local area cooling versus broad area cooling for boil-off reduction in large-scale liquid hydrogen storage tankspublishedVersio
Rock Mass Classification in TBM Tunneling using Artificial Neural Network Techniques: A Case Study from Siwalik Region of Nepal
Full text linkThe Himalayan region exhibits a highly complex geological setting influenced by tectonic activities, resulting in faulted, folded, sheared, and deeply weathered rock masses. This study focuses on the Siwalik region of Nepal, where the geology is characterized by sandstone, mudstone, siltstone, and clay. Accurate rock mass characterization is crucial for Tunnel Boring Machine (TBM) tunneling projects, particularly in the challenging geological conditions of the Nepal Himalayas. Empirical rock mass classification systems, such as the Rock Mass Rating (RMR) and Q-system, often fall short in TBM operations due to limited access to the tunnel face and the dynamic nature of TBM excavation. To address these challenges, this research employs a machine learning (ML) technique to classify rock mass conditions using operational and geological data collected from the Sunkoshi Marin Diversion Multipurpose (SMDM) project in Nepal, where a double-shield TBM was used to excavate the 13.3 km long headrace tunnel. A comprehensive dataset comprising 3,173 TBM cycles, including parameters such as cutter head speed, torque, thrust, and penetration rates, was utilized for model development. An Artificial Neural Network (ANN) model was developed, trained, and optimized using grid search to identify the best hyperparameters. The Synthetic Minority Oversampling Technique (SMOTE) was applied to address the class imbalance, significantly improving the model's recall for Class V (poor rock mass class). Performance metrics such as accuracy, precision, recall, and F1-score were used to evaluate the model. Additionally, SHAP (Shapley Additive Explanations) analysis was conducted to interpret feature contributions for rock mass Class V, which revealed that torque and thrust had the highest influence on predicting poor rock mass conditions. This study highlights the effectiveness of ML models in improving rock mass classification, especially for underrepresented classes, and provides valuable insights for optimizing TBM operations in complex geological settings.publishedVersio
Synthesis and Electrochemical Evaluation of PTMA-Based Cathode Materials for Sustainable Organic Radical Batteries
Full text linkDenne avhandlingen undersøker poly(2,2,6,6-tetrametyl-1-piperidinyloxy-4-yl metakrylat) (PTMA) som katodemateriale for miljøvennlige og høytytende organiske batterier, med sikte på å forbedre syklusstabiliteten og kapasitetsopprettholdelsen og evaluere det som en egnet erstatning for konvensjonelle litiumionebatterier. PTMA ble syntetisert ved hjelp av ulike polymeriseringsmetoder, inkludert fri radikalpolymerisering og gruppeoverføringspolymerisering med og uten tverrbinding, og testet i litiumhalvceller under ulike strømtettheter. Den tverrbundne PTMA-en viste en reversibel spesifikk ladning på ca. 103 Ah kg -1 og beholdt over 95 % av kapasiteten etter 200 lade- og utladningssykluser ved 2 C, i motsetning til de ikke-tverrbundne variantene som hadde et betydelig kapasitetstap tidlig i prosessen. Disse resultatene viser at egenskaper som reduserer polymerens løselighet, spesielt tverrbinding, er avgjørende for å oppnå både høy kapasitet og langsiktig stabilitet i PTMA-baserte batterier.This thesis investigates poly(2,2,6,6-tetramethyl-1-piperidinyloxy-4-yl methacrylate) (PTMA) as a cathode material for environmentally friendly and high-performance organic batteries, aiming to improve cycling stability and capacity retention and evaluate it as a suitable substitute to conventional lithium-ion batteries. PTMA was synthesized using different polymerization methods, including free radical polymerization and group-transfer polymerization with and without cross-linking, and tested in lithium half-cells under various current densities. The cross-linked PTMA showed a reversible specific charge of approximately 103 Ah kg -1 and retained over 95 % of its capacity after 200 charge-discharge cycles at 2 C, in contrast to non-cross-linked variants which suffered significant early capacity loss. These results demonstrate that properties reducing polymer solubility, particularly cross-linking, is critical for achieving both high capacity and long-term stability in PTMA-based batteries
Application of Koopman Operator Theory for Control of an Autonomous Formula Student Race Car
No full textDenne oppgaven undersøker anvendelse av Koopman operator-teori for modellering og prediktiv styring av en autonom Formula Student-racerbil. Med Koopman operator-teori er det mulig å transformere ulineære dynamiske systemer til løftede lineære dynamiske systemer ved å betrakte dynamikken til målinger av systemtilstanden. Dette åpner opp for å kunne bruke standardmetoder for styring av lineære systemer til å styre ulineære systemer. Denne oppgaven benytter teknikker basert på dynamisk modus dekomponering og dyp læring for å konstruere løftede modeller av en enkel ikke-afin ulineær bilmodell basert på en sykkelmodell. Nøyaktigheten til de konstruerte løftede modellene er sammenliknet med både den ekte modellen og en modell basert på lokal linearisering. Videre er modellene brukt i lineær modellprediktiv regulering for å simulere oppførselen i lukket sløyfe. Blant de testede metodene viste det seg at løfteteknikken med dyp læring ga den beste løftede lineære modellen. Den ulineære bilmodellen kunne læres godt nok til at prediktiv regulering kunne benyttes til å kjøre rundt en bane i simulering, men modellen var ikke kapabel til å fremprovosere hjørnekuttende oppførsel ved parameterjustering. Et forsøk med en mer avansert bilmodell viste seg å være for vanskelig å lære til at prediktiv regulering kunne benyttes med den lærte modellen.This thesis investigates application of Koopman operator theory for modeling and predictive control of an autonomous Formula Student race car. With Koopman operator theory it is possible to transform nonlinear dynamical systems into lifted linear dynamical systems by considering the evolution of measurements of the state. This has the desirable potential of allowing standard control methods for linear systems to be used to control nonlinear systems. This thesis utilizes techniques based on dynamic mode decomposition and deep learning to construct such lifted models of a simple non-affine nonlinear vehicle model based on a bicycle model. The identified models' accuracy are compared against the ground truth model and a baseline model obtained using local linearization. Furthermore, the lifted Koopman models are used in linear model predictive control to simulate closed-loop behavior. Among the methods tested, the experiments found that a neural network-based lifting yielded the best lifted model, and the vehicle model could be learned well enough with the implemented methods to perform predictive control to drive around a race track in simulation. The best lifted model was however incapable of providing corner-cutting behavior through tuning. An experiment with a higher-fidelity vehicle model also proved too difficult to learn to be viable for predictive control