|
|
Non destructive sensors for measurement of the very slow change of the biological structure dimensions
Kostka, Filip ; Kubásek, Radek (referee) ; Friedl, Martin (advisor)
The aim of this thesis is to record a very small increment of the perimeter trees in one day using dendrometr. As the sensor is used tensometer strain gauge, which the deformation of its shape also changes its resistance and it is used subsequently to monitor the recording of growth. Data from dendrometr are handles by digital multimeter, which is synchronized with the PC, where further processing occurs date. This semester project also deals with the thermal expansion of metal parts of dendrometr and ambient conditions during the experiment.
|
|
|
Artificial neural network for modeling electromagnetic fields in a car
Kostka, Filip ; Škvor, Zbyněk (referee) ; Raida, Zbyněk (advisor)
The project deals with artificial neural networks. After designing and debugging the test data set and the training sample set, we created a multilayer perceptron network in the Neural NetworkToolbox (NNT) of Matlab. When creating networks, we used different training algorithms and algorithms improving the generalization of the network. When creating a radial basis network, we did not use the NNT, but a specific source code in Matlab was written. Functionality of neural networks was tested on simple training and testing patterns. Realistic training data were obtained by the simulation of twelve monoconic antennas operating in the frequency range from 2 to 6 GHz. Antennas were located inside a mathematical model of Octavia II. Using CST simulations, electromagnetic fields in a car were obtained. Trained networks are described by regressive characteristics andthe mean square error of training. Algorithms improving generalization are applied on the created and trained networks. The performance of individual networks is mutually compared.
|
|
|
Dynamic models of earthquake source and modeling of seismicity
Kostka, Filip ; Gallovič, František (advisor) ; Matyska, Ctirad (referee)
In the present thesis we perform modeling of earthquake source using laboratory derive rate-and-state laws of friction. We have developed a code in Fortran 90 for modeling a planar, two-dimensional fault with general dip and heterogeneous distribution of frictional parameters. We use a quasi-dynamic approximation and assume that the fault is submnerged in an infinite elastic half-space. We performed an extensive number of numerical experiments to study the effect of fricitonal parameters distribution on the spatio-temporal complexity of slip on fault. We also study the effect of the so called Coulomb stress changed on clock advance and clock delay of events. For this purpose we use both a homogeneous model and a model of random frictional parameteres which exhibits the Gutenberg-Richter frequency- size dependence in the range of two magnitudes. We find that the effect of Couloumb stress change is nontrivial and depends on factors such as the domain of stress load and the slip velocity on it. Powered by TCPDF (www.tcpdf.org)
|
|
|
Dynamic models of the earthquake source
Kostka, Filip ; Gallovič, František (advisor) ; Burjánek, Jan (referee) ; Kaneko, Yoshihiro (referee)
Dynamic models of the earthquake source allow simulating the evolution of stress and slip at tectonic faults by coupling the equations of motion in a volume surrounding the fault with a constitutive law that represents the surface forces acting on the fault. In the first part of the thesis, we review important properties of shear rupture for brittle, linear slip-weakening, and rate-and-state constitutive laws. In the second part of the thesis, we present two studies utilizing 3-D dynamic modeling at both long (hundreds of years) and short (seconds) time scales. In the first study, we model seismic cycles using the rate- and-state laws of friction and perform a parametric exploration of the effects of sudden intra-cycle shear stress perturbations on the clock advance or delay of the subsequent large event. We find that when the perturbation is applied during specific time intervals, the earthquakes following the perturbation are only small ruptures that do not completely release stress on the whole fault. The time interval between large earthquakes may thus be prolonged up to 80% when compared to the unperturbed cycles. We reproduce this behavior on a numerical heterogeneous model of the Parkfield segment of the San Andreas fault and demonstrate that the mechanism could have been responsible for the observed large...
|
|
|
Dynamic models of earthquake source and modeling of seismicity
Kostka, Filip ; Gallovič, František (advisor) ; Matyska, Ctirad (referee)
In the present thesis we perform modeling of earthquake source using laboratory derive rate-and-state laws of friction. We have developed a code in Fortran 90 for modeling a planar, two-dimensional fault with general dip and heterogeneous distribution of frictional parameters. We use a quasi-dynamic approximation and assume that the fault is submnerged in an infinite elastic half-space. We performed an extensive number of numerical experiments to study the effect of fricitonal parameters distribution on the spatio-temporal complexity of slip on fault. We also study the effect of the so called Coulomb stress changed on clock advance and clock delay of events. For this purpose we use both a homogeneous model and a model of random frictional parameteres which exhibits the Gutenberg-Richter frequency- size dependence in the range of two magnitudes. We find that the effect of Couloumb stress change is nontrivial and depends on factors such as the domain of stress load and the slip velocity on it. Powered by TCPDF (www.tcpdf.org)
|
|
|
Quasidynamic modeling of tectonic faults: segmentation of earthquakes
Kostka, Filip ; Gallovič, František (advisor) ; Souček, Ondřej (referee)
Some tectonic faults are documented to release strain in form of segmented earthquakes, when only a part of the fault fails at a given time, being followed by a rupture of another part of the fault years to tens of years later. In this thesis, our aim is to explain such behaviour using an idealized dynamic-fault model governed by a laboratory derived friction law. To this end, we rst give an overview of a class of rate- and-state friction (RSF) laws, commonly used in modelling of seismic cycles. We perform numerical tests on a spring slider model using the most recent version of the RSF law proposed by Nagata et al. (2012) to better understand its features. Performing extensive numerical experiments with velocity- strenghtening and velocity-weakening zones of varying sizes along an in nite two-dimensional vertical fault, we nd examples of the earthquake segmentation. Powered by TCPDF (www.tcpdf.org)
|
|
|
Non destructive sensors for measurement of the very slow change of the biological structure dimensions
Kostka, Filip ; Kubásek, Radek (referee) ; Friedl, Martin (advisor)
The aim of this thesis is to record a very small increment of the perimeter trees in one day using dendrometr. As the sensor is used tensometer strain gauge, which the deformation of its shape also changes its resistance and it is used subsequently to monitor the recording of growth. Data from dendrometr are handles by digital multimeter, which is synchronized with the PC, where further processing occurs date. This semester project also deals with the thermal expansion of metal parts of dendrometr and ambient conditions during the experiment.
|
|
|
Artificial neural network for modeling electromagnetic fields in a car
Kostka, Filip ; Škvor, Zbyněk (referee) ; Raida, Zbyněk (advisor)
The project deals with artificial neural networks. After designing and debugging the test data set and the training sample set, we created a multilayer perceptron network in the Neural NetworkToolbox (NNT) of Matlab. When creating networks, we used different training algorithms and algorithms improving the generalization of the network. When creating a radial basis network, we did not use the NNT, but a specific source code in Matlab was written. Functionality of neural networks was tested on simple training and testing patterns. Realistic training data were obtained by the simulation of twelve monoconic antennas operating in the frequency range from 2 to 6 GHz. Antennas were located inside a mathematical model of Octavia II. Using CST simulations, electromagnetic fields in a car were obtained. Trained networks are described by regressive characteristics andthe mean square error of training. Algorithms improving generalization are applied on the created and trained networks. The performance of individual networks is mutually compared.
|
guest ::
RSS feed.