Explore CTU's Digital Library

Access research publications, theses and academic works

Faculty of Civil Engineering Faculty of Mechanical Engineering Faculty of Electrical Engineering Faculty of Nuclear Sciences and Physical Engineering Faculty of Architecture Faculty of Transportation Sciences Faculty of Biomedical Engineering Faculty of Information Technology Klokner Institute Masaryk Institute of Advanced Studies Central Library Acta Polytechnica Acta Polytechnica CTU Proceedings CTU Publishing Activity

Recent Submissions

Environmental qualification of a high-temperature superconducting magnet and flux pump system for spaceflight
(IOP Publishing, 2026) Pavri B.; Pollock R.; Long N.; Goddard-Winchester M.; Malich M.; Filgas R.
The Heki technology demonstration payload was launched to the International Space Station (ISS) in September 2025. Heki is designed to advance the technology readiness level of two novel technologies: a high-temperature superconducting magnet and a 'flux pump'-an inductive superconducting rectifier which energizes the magnet to field. This paper describes the rigorous pre-launch test and analysis program developed to meet ISS safety requirements and to demonstrate that these new components will survive the environment of launch and operations in space. The tests included structural analyses and random vibration testing to emulate launch conditions, thermal vacuum testing to simulate operations in space, and electromagnetic compatibility testing to verify Heki can be operated without interfering with neighbouring experiments and electronics. Heki successfully completed this pre-launch test program without degradation in performance, reducing the risk of failure in flight and marking an important milestone toward commercialization of these technologies for space applications.
Analysis of mixed radiation fields at the MoEDAL experiment based on real-time data from a Timepix detector network
(Springer, 2026) Bergmann B.; Burian P.; Janeček J.; Leroy C.; Mánek P.; Pinfold J.; Pospíšil S.; Soluk R.; Suk M.
The primary objective of this work is the determination of fluences and characteristics of fast neutrons, other hadrons, and highly ionizing particles (HIPs) in the environment of the MoEDAL experiment at the Large Hadron Collider. These particles may constitute an experimental background for the passive Nuclear Track Detectors (NTDs) used by MoEDAL to search for tracks potentially produced by magnetic monopoles, in particular by particles indistinguishable in NTD from monopoles. The study is based on data acquired by the Timepix hybrid silicon pixel detector network, which represents the first and only active detector system installed and operated as part of the MoEDAL experiment from 2013 to 2018. The Timepix detector network enables real-time measurements of mixed radiation fields, including the composition, spectral properties, and directional characteristics of individual radiation components across different regions of the MoEDAL experimental area. The paper presents detailed results of the radiation field analysis with emphasis on neutrons and HIPs, including their directional distributions. The first results demonstrating the spatial tracking capabilities of the Timepix detectors are also reported, illustrating the reconstruction of particle direction and energy-loss profiles from individual detector frames.
Performance Study of Compact Semiconductor Neutron Spectrometer HardPix for Lunar Water Mapping
(Multidisciplinary Digital Publishing Institute (MDPI AG), 2026) Filgas R.; Matthiä D.; Cintas H.; Slavíček T.; Jelínek J.; Gohl S.; Malich M.; Ferreira Natal Da Luz P.; Bergmann B.
The current interest in lunar exploration led by the Artemis program is pushing scientists to search for lunar water deposits directly on the surface of the Moon, using small robotic rovers. The Institute of Experimental and Applied Physics, Czech Technical University, Prague (IEAP CTU), is developing a miniature Timepix3-based detector called Neutron HardPix, which is capable of mapping water deposits using non-invasive detection of neutrons created underground by cosmic rays and thermalized by hydrogen. This neutron spectrometer measures count rate variations in thermal, epithermal and fast neutrons attributed to hydrogen abundance in the lunar subsurface, while monitoring cosmic radiation as a natural source of neutrons. Neutron HardPix is based on the miniature (<0.1 U, 130 g) radiation monitor HardPix, and has significant space heritage.
Analysis of Weak Signals in Images
Analýza slabých signálů v obrazech
(Czech Technical University in Prague) Špetlík, Radim; Matas, Jiří; Kristan, Matej; Zhao, Guoying
The challenge of modern computer vision has shifted from the analysis of clearly observable phenomena to the recovery of "weak signals" - information that is fundamentally corrupted by sub-perceptual magnitude, temporal integration, chemical variance, or semantic abstraction. This dissertation argues that ill-posed inverse problems in these regimes are effectively addressed by a unified methodology: transforming confounded evidence into robust visual representations and constraining their interpretation with strong, data-driven generative priors.This thesis validates this hypothesis through four distinct case studies. First, in the domain of remote physiological monitoring, we address the recovery of the sub-perceptual pulse signal from video. We introduce HR-CNN, an end-to-end learned method that replaces traditional signal processing with a representation optimized for a low Signal-to-Noise Ratio (SNR). We demonstrate that this approach implicitly learns to isolate skin regions and disentangle the physiological signal from severe head motion, significantly outperforming analytic baselines on a novel challenging dataset.Second, in the domain of high-velocity dynamics, we introduce SI-DDPM-FMO, a method for recovering the trajectory and shape of Fast Moving Objects (FMOs) from a single blurred image. By leveraging a Denoising Diffusion Probabilistic Model (DDPM) as a generative prior, we effectively perform temporal super-resolution, synthesizing the missing temporal sequence without the extrinsic redundancy required by other methods.Third, we address the domain of olfactory biometrics, treating the high-dimensional, drifting output of chemical sensors (GCxGC-TOF-MS) as multi-channel images. We present the Human Scent Dataset, the largest of its kind, and demonstrate that Convolutional Neural Networks (CNNs) can learn invariance to instrumental drift, establishing the feasibility of identity verification from raw scent without explicit registration.Fourth, we tackle the loss of structural identity in generative video stylization. We introduce StructuReiser, a framework that utilizes a pre-trained diffusion model not as a generator, but as a structural critic. By projecting stylized frames onto a manifold constrained by this structural prior, we achieve real-time, temporally consistent video synthesis that preserves the subject's identity against the noise of artistic abstraction.Collectively, this work demonstrates that the boundary between signal and noise is not an immutable property of the sensor, but a function of the representation and the prior. By moving from the direct reconstruction of noisy observations to the projection of weak signals onto learned manifolds, we expand the operational envelope of computer vision into domains of physiological invisibility, high temporal velocity, chemical complexity, and generative ambiguity.
Validation of Computational Software for Criticality Safety Analysis of Spent Nuclear Fuel Systems
Validace výpočetních nástrojů pro výpočty kritičnosti systémů s vyhořelým jaderným palivem
(Czech Technical University in Prague) Šikl, Matěj; Vočka, Radim; Vrban, Branislav; Lovecký, Martin
The thesis focuses on the subcriticality assessment of spent nuclear fuel storage systems using the burnup credit methodology as an alternative to the traditionally conservative fresh-fuel approach. To validate the computational codes, a database of Czech nuclear power plant start-up tests (reactor criticals) was compiled, and their similarity to storage systems was evaluated using the Serpent 2 and SCALE (TSUNAMI-IP) code systems. The results demonstrated a high similarity between reactor criticals and spent fuel pools, whereas cask storage systems were found to be well represented by specific fresh-fuel experiments. Based on these findings, a new subcriticality assessment methodology including uncertainty analysis was developed and successfully verified through pilot applications, demonstrating a significant increase in calculated safety margins.