51 research outputs found
Gesteinsklassifikation im Tunnelbau basierend auf seismischen Geschwindigkeiten und Vortriebsparametern unter Verwendung von Support-Vektor- Maschinen
No full textAbstract I Zusammenfassung III 1 Introduction 1 1.1 Motivation 2 1.2 Aim 3 1.3
Selection of a Self-Learning Algorithm 4 1.4 Geological Rock-Mass Properties 8
Calculation of RMR and RQD 9 1.5 Tunnel-Driving Data 15 1.5.1 Operating Mode
of an Open Gripper Tunnel-Boring Machine 15 1.6 Relationships between Rock
Quality, Seismic Velocities and Tunnel-Driving Parameters 18 1.6.1 Seismic
Velocities and Rock-Mass Behavior 18 1.6.2 Tunnel-Driving Parameters and Rock-
Mass Behavior 23 1.7 Seismic Systems in Tunneling 24 1.8 Basic Concept of
Support Vector Machines 27 2 Field Investigations 2.1 Geological Setting 35
2.1.1 The Faido Adit 35 2.1.2 Geological Setting of the Glendoe Tunnel 41 2.2
Seismic Measurements 48 2.2.1 Seismic Data Acquisition in the Faido Adit 48
2.2.2 Seismic Data Acquisition in the Glendoe Tunnel 49 2.3 Acquisition of
Tunnel-Driving Data in the Glendoe Tunnel 54 3 Statistical Evaluation 3.1
Statistical Evaluation of the Faido Data Set 57 3.2 Statistical Evaluation of
the Glendoe Data Set 62 3.2.1 Final Remarks 69 4 Development of a SVM for
Rock-Mass Classification 71 4.1 Programming Environment 81 5 Results of Rock-
Mass Classification using SVMs 83 5.1 RQD Prediction on the Faido Adit Data
Set 84 5.2 RMR Prediction on the Faido Adit Data Set 90 5.3 RQD Prediction in
the Glendoe Tunnel 1-m Data Set 93 5.4 RMR Prediction on the Glendoe Tunnel
1-m Data Set 96 5.5 RQD Prediction on the Glendoe Tunnel 4-m Data Set 101 5.6
RMR Prediction on the Glendoe Tunnel 4-m Data Set 103 6 Discussion 107 7
Conclusion 115 8 Outlook 117 Acknowledgements 119 References 121 List of
Figures 132 List of Tables 136 Appendices 138 A Nomenclature 141 B Field
Surveys 145 C Statistics 147 Curriculum Vitae 154The continuously increasing demand on safe and cost-efficient tunnel
constructions worldwide has led to the development of seismic systems as a
predictive tool ahead of tunneling. These seismic systems are either
specialized for hard rock or soft rock excavations. In this study, the
hypothesis is tested, if artificial intelligence approaches are capable to
deduce automatically and in real time critical rock parameters out of seismic
observations. The hypothesis is tested in hard rock environments, using
combined geological and seismic observations of the Faido Adit (Gotthard Base
Tunnel - Switzerland) and the Glendoe Tunnel (Hydro Electric Power Plant near
Loch Ness - Scotland UK). The evaluation of geotechnical rock-mass behavior in
hard rock is commonly based on a rock-mass classification. Especially, a fast
assessment of the hard rocks´ bearing capacity is mandatory to quickly
ascertain the required structural tunnel support. Therefore, a user-oriented
geotechnical interpretation of seismic results in real time is tested. Within
this study, a support vector machine (SVM) is applied to the discovery and
automated prediction of relationships between seismic P- and S-wave velocities
with heuristic rock-mass classification systems, such as the widely used Rock
Quality Designation (RQD) index or the Rock Mass Rating (RMR) factor. The data
available for this task were acquired during two field surveys in hard rock
using the Integrated Seismic Imaging System ISIS and geotechnical mapping of
the rock mass. The first survey was carried out in the gneisses of the Faido
Adit, which is part of the Gotthard Base Tunnel in Switzerland. Seismic
velocity data from a 2-D tomography with a cells size of 0.5 m in direction of
the excavation along a 448 m long seismic profile have been used. The second
seismic survey took place along a 300-m long profile in quartz schists and
quartz-mica schists in the headrace tunnel of the Glendoe Hydro Electric Power
Plant in the Scottish Highlands. The Glendoe Tunnel was excavated with a
tunnel-boring machine (TBM), such that adaptations had to be made to the
seismic setup of the TBM-integrated seismic measurements. These adaptations
let to a 1-D tomography with a cell size in tunnel direction of 4 m and
therefore, to a considerably reduced resolution in the available seismic
velocity data, compared to the Faido Adit data set. Thus, the SVM approach was
applied separately to the two data sets. As there may exist some direct or
indirect link between rock-mass classes and tunnel-driving parameters, such as
the thrust force, the penetration rate, the cutter-head torque and the cutter-
head speed, these properties were included in the data base of the Glendoe
Tunnel survey. The tunnel-driving parameters exhibit a much higher spatial
resolution than the seismic data, such that their information content was
first explored by training and testing a SVM solely on this data with a
resolution of 1 m. In both data sets, 3 RQD classes and 2 RMR classes were
distinguished. Two fundamentally different results are achieved during rock-
mass classification based on the data sets from the Faido Adit and the Glendoe
Tunnel: 1\. Based on high-resolution seismic data from the Faido Adit, the
classification of RQD or RMR classes proofed feasible. 2\. Based on either
tunnel-driving data, or else tunnel-driving and seismic data combined, from
the Glendoe Tunnel with lower resolution, the RQD and RMR classification did
not provide satisfying results. The variability in the rock-mass quality,
expressed either as RQD or RMR, is extremely low for the Faido Adit, leading
to strong proximity of most data samples to the class boundaries. The
detection of patterns that link the rock-mass classes to the seismic
velocities in the Faido Adit data set is therefore remarkable, especially for
the small number of training samples available and despite a strong tendency
to overfit. For the Glendoe Tunnel, the training and testing of the SVM
reveals that the classes were not or poorly classified by the automated
classification approach. The models based exclusively on tunnel-driving
parameters show severe cases of overfitting and extremely low generalization
ability. These results do not rule out that higher order correlations exist
between tunnel-driving parameters and rock-mass classes in general, but no
evidence on this has been discovered in this study. The additional use of
seismic body-wave velocities in the Glendoe Tunnel has been inevitably
accompanied by a significant reduction of the data set. Adding the seismic
velocities to the data base did not influence the classification result
positively. This let to the assumption that the data set is by far too small
for a proper learning process, such that no rules were learned from the data
set and the prediction failed in consequence. The quality and spatial
resolution of the seismic observations is therefore crucial for the
reliability of the prediction of rock-mass classes. The quality and cell size
of the underlying seismic tomography strongly depends on the seismic layout
during the data acquisition, such that the careful planning of the seismic
survey can be determined as a key requirement for the success of a fast and
automated rock-mass classification and the detection of hazardous zones in the
rock mass. Nevertheless, even with the limited size of the available data
sets, it was possible to show that SVMs are a powerful tool in real time
expert systems for geotechnical applications. It has been proven within this
study that it is possible to predict rock-mass classes out of high resolution
seismic data with high accuracy.Die Entwicklung speziell auf den Tunnelbau abgestimmter seismischer Systeme
zielt darauf ab, den stetig wachsenden Ansprüchen an die Sicherheit im
Tunnelbau, bei gleichzeitiger Kostenreduktion, gerecht zu werden. Diese
seismischen Methoden sind auf die speziellen Anforderungen im Hart- oder
Lockergestein angepasst. In der vorliegenden Arbeit wird die Hypothese
getestet, dass Methoden der Künstlichen Intelligenz genutzt werden können, um
automatisiert und zeitnah kritische Gesteinsparameter aus seismischen
Beobachtungen abzuleiten. Zur Überprüfung dieser Hypothese wurden seismische
und geologische Daten aus zwei Feldeinsätzen in Hartgestein verwendet, und
zwar aus dem Faido Zugangsstollen (Gotthard Basis Tunnel - Schweiz) und dem
Glendoe Tunnel (Wasserkraftwerk am Loch Ness - Schottland). Besonders im
Hartgestein erfolgt eine zeitnahe Einteilung der geotechnisch relevanten
Eigenschaften meist über Systeme zur Gesteinsklassifikation. Die geotechnische
Klassifikation von Gesteinen dient vor allem der Abschätzung der
Standfestigkeit des Gebirges vor Ort, als wichtiger Voraussetzung zur
Ermittlung des nötigen Ausbaus und damit für die Stabilität und Sicherheit des
Tunnels. Da eine umfassende, anwendungsorientierte und zeitnahe, geotechnische
Interpretation der im Hartgestein gewonnenen seismischen Daten bisher nicht
gewährleistet ist, wurde in der vorliegenden Arbeit eine auf seismischen Daten
basierende Routine zur automatischen und zeitnahen geotechnischen
Gesteinsklassifikation mit Support Vektor Maschinen (SVMs) entwickelt. Der
Ansatz wurde auf zwei verbreitete Systeme zur Gesteinsklassifikation
angewendet: den Rock Quality Designation (RQD) Index und den Rock Mass Rating
(RMR) Faktor. Datensätze aus zwei Feldeinsätzen im Hartgestein, die mit dem
Integrated Seismic Imaging System ISIS durchgeführt wurden, standen hierfür
zur Verfügung. Der erste Feldeinsatz erfolgte über eine Profillänge von 448 m
in den Gneisen des, im Sprengvortrieb errichteten, Faido Zugangsstollens zum
Gotthard Basis Tunnel (südliche Schweiz). Die seismischen Geschwindigkeiten
entlang des Profils im Faido Zugangsstollens basieren auf einer 2D-Tomographie
mit einer Zellgröße entlang der Tunnelachse von 0.5 m. Der zweite Feldeinsatz
wurde in Quarzschiefern und Quarzglimmerschiefern entlang eines 300 m langen
Profils im Triebwassertunnel des Wasserkraftwerkes Glendoe (schottisches
Hochland) durchgeführt. Der Glendoe Tunnel wurde maschinell vorgetriebenen,
wodurch entsprechende Anpassungen in der Geometrie der seismischen
Datenakquisition nötig wurden. Diese Anpassungen führten zu einer verringerten
Qualität und Auflösung der seismischen Daten, woraus eine 1D-Tomographie mit 4
m Zellgröße entlang der Tunnelachse berechnet wurde. Die Datensätze des Faido
Stollens und des Glendoe Tunnels wurden daher getrennt evaluiert. Es wurde ein
Zusammenhang zwischen Vortriebsparametern der Tunnelbohrmaschine, wie
Vortriebspressenkraft, Penetration sowie Drehmoment und Drehzahl des
Schneidrads, mit dem RQD und RMR angenommen. Die Vortriebsparameter wurden
daher in den Ansatz zur automatischen Vorhersage von Gesteinsklassen
integriert. Da die Vortriebsparameter eine sehr viel höhere Auflösung als die
seismischen Daten besitzen, wurden diese über 1 m große Intervalle gemittelt
und zuerst separat betrachtet. In beiden Feldstudien wurden drei RQD-Klassen,
sowie zwei RMR-Klassen unterschieden. Zwei grundsätzlich verschiedene
Ergebnisse wurden für die Gesteinsklassifikationen auf Grundlage der Daten aus
dem Faido Zugangstollen oder dem Glendoe Tunnel erreicht: 1\. Basierend auf
den höher aufgelösten seismischen Daten des Faido Zugangsstollens konnte eine
erfolgreiche Gesteinsklassifikation sowohl für die RQD als auch für die RMR-
Klassen vorgenommen werden. 2\. Basierend auf den niedriger aufgelösten Daten
des Glendoe Tunnels konnten keine zufriedenstellenden
Klassifikationsergebnisse erreicht werden. Dies gilt sowohl für eine getrennte
Betrachtung von Vortriebsparametern mit einer höheren Auflösung von 1 m, als
auch für den kombinierten Datensatz aus seismischen Daten und
Vortriebsparametern mit einer Auflösung von 4 m. Die Variabilität des RQD und
RMR im Datensatz des Faido Stollens ist gering. Die erfolgreiche
Klassifikation ist daher, insbesondere trotz der geringen Anzahl von zur
Verfügung stehenden Datenpunkten und einer deutlichen Tendenz des SVM-Models
hin zu Überanpassung an die Trainingsdaten, bemerkenswert. Eine Analyse der
Ergebnisse zum Glendoe Tunnel zeigte, dass die Klassen nicht oder sehr
schlecht klassifiziert wurden. Die SVM-Modelle der RQD und RMR Klassifikation,
die ausschließlich auf Vortriebsparametern basieren, zeigten extreme Anpassung
an die Trainingsdaten und geringe Generalisationsfähigkeit. Diese Ergebnisse
schließen zwar nicht aus, dass generell ein Zusammenhang zwischen
Vortriebsparametern und Gesteinsklassen bestehen kann, in dieser Arbeit konnte
dies jedoch nicht verifiziert werden. Der Einbezug der seismischen
Geschwindigkeiten, mit einhergehender Reduktion der Datensatzgröße, ergab
keine positive Beeinflussung des Ergebnisses. Dies lässt den Schluss zu, dass
der Datensatz eine zu geringe Anzahl und Qualität an Datenpunkten aufweist, so
dass keine Regeln für die Klassifikation aus den Daten abgeleitet werden
konnten und eine Vorhersage in der Konsequenz nicht möglich ist. Die Qualität
und räumliche Auflösung der Tomographie ist daher entscheidend für die
Aussagekraft einer Vorhersage von Gesteinsklassen. Dies hängt stark von der
Anordnung der Quellen und Empfänger während der seismischen Datenakquisition
ab. Eine umsichtige Planung der Datenakquisition ist daher unerlässliche
wichtige Voraussetzung für eine erfolgreiche automatisierte
Gesteinsklassifikation. Trotz einer stark limitierten Größe der zur Verfügung
stehenden Datensätze konnte gezeigt werden, dass SVMs als mächtiges Werkzeug
in einem Expertensystem für geotechnische Fragestellungen genutzt werden
können. Es konnte in dieser Arbeit gezeigt werden, dass eine genaue Vorhersage
von Gesteinsklassen, basierend auf hochauflösenden seismischen Messungen,
möglich ist
Nachhaltigkeitsaspekte der Urangewinnung – Anmerkungen zur Taxonomie-Entscheidung der EU
No full textKernkraft ist im Rahmen der EU-Taxonomie zu einer grünen Technologie erklärt worden, ohne jedoch bei dieser Entscheidung die bergbauliche Gewinnung des Urans und die Endlagerung zu berücksichtigen. Deshalb werden in der vorliegenden Arbeit die verschiedenen Urangewinnungsmethoden und ihre Wirkungen auf die Umwelt dargelegt. Weil es sich um den Abbau eines Gefahrstoffes handelt, entstehen zusätzlich zu den üblichen Beeinträchtigungen durch Bergbau spezifische Risiken und zukünftige Lasten. Auch die Aufbereitung und Weiterverarbeitung ist risikoreich und belastet Umwelt und Menschen. Es besteht ein signifikantes Missverhältnis zwischen den oft sozialisierten Sanierungskosten solcher Altlasten und den Gewinnen aus dem Uranbergbau. Das zweite Kapitel bewertet diese Fakten und zeigt auf, dass Umwelt-, Sozial- und Governancebelange des Uranbergbaus sehr hohe bis hohe Risiken bergen, die v. a. in der Radioaktivität und Toxizität der Rohstoffe gründen. Hinzu kommen
politische Risiken durch die enge Beziehung zwischen der Nutzung des Urans als Energiequelle und seiner Verwendung für die atomare Rüstung
Operational design and implementation experience with due diligence initiatives and certification schemes
No full textThis policy brief analyses the research results on success factors and challenges in the application of due diligence and certification schemes to meet socio-economic demands and establish chains of custody in the mining sector
GPU-accelerated CFD Simulations for Turbomachinery Design Optimization
No full textDesign optimization relies heavily on time-consuming simulations, especially when using gradient-free optimization methods. These methods require a large number of simulations in order to get a remarkable improvement over reference designs, which are nowadays based on the accumulated engineering knowledge already quite optimal.High-Performance Computing (HPC) is essential to reduce the execution timeof the simulations. While parallel programming using the CPU is established since more than two decades, the use of accelerators, such as the Graphics Processing Unit (GPU), is relatively recent in design optimization. The GPU has actually a huge computational power comparable to a many-core cluster but concentrated in one device. This raw power is not easy to utilize as entire code parts have to be rewritten using a GPU programming language. Even though high-level standards (e.g. openACC) are able to bring a basic acceleration with a low development effort, it is not simple to get large speedups with these methods. Low-level programming languages are more efficient but different speedups are reported and there is a need fora deep analysis to make the GPU potential more transparent to scientists especially non-experts in HPC.In order to study the GPU acceleration for CFD steady simulations, two in-house CFD solvers have been ported to the GPU; one with explicit and the second with implicit time-stepping. After the porting and the validation of the GPU solvers, the GPU code optimization leads to the identification of a set of key parameters affecting the GPU efficiency. At the same time, both methods have been compared resulting into a performance model and a classification of the GPU acceleration of some CFD operations. The purpose is to enable scientists to take an educated decision concerning the GPU porting of their CPU applications by providing an expected GPU speedup.In addition to the two GPU CFD solvers that are now integrated into the in-house design optimization software package, this research provided key elements to reduce the ambiguity about the GPU potential, namely a qualitative analysis and a classification. These tools can help selecting the best candidate for a breakthrough in CFD acceleration. At the same time, this work identified serious limitations in the preconditioning of a linear system of equations and the limit of today iterative matrix factorization methods in terms of stability and convergence. There is a need for a paradigm shift toward inherently parallel preconditioners. The developed tools have been used for the optimization of a compressor and a turbine cascade resultinginto a faster optimization process on the GPU
Usability of Rich Internet Applications
No full textThis thesis reports the outcome of a study that is performed for Info Support. Info Support is a software house that is interested in new technologies. Unit finance and insurance is interested in the usability of Rich Internet Applications. Rich Internet Applications are new kind of web applications that offer more possibilities to design user friendly and attractive graphical user interface (GUI) that is comparable to the GUI of desktop applications. To investigate if the usability of the Rich Internet Application is indeed better then the traditional web application, two car insurance web applications are built as a proof of concept. These web applications are built with the same usability principles and have the same content and functionalities. The only difference is the technology that is used for building the web applications. The RIA is built with JavaFX and the traditional web application with PHP/HTML. These two applications are tested in an experiment where 26 participants with different ages and sex have tested two web applications in a lab. The results of this experiment is statistically analyzed to find dependency. From the analysis we have found that the satisfaction is significant for the tested web applications and therefore we can conclude that the usability of the Rich Internet Applications is indeed more usable than the traditional web application. This finding can be useful for finance and insurance companies to convert their web applications into RIA to attract more clients to buy their products.Media knowledge engineeringElectrical Engineering, Mathematics and Computer Scienc
Well-posedness and energy decay for Timoshenko systems with discrete time delay under frictional damping and/or infinite memory in the displacement
No full textInternational audienceIn this paper, we consider a vibrating system of Timoshenko-type in a bounded one-dimensional domain with discrete time delay and complementary frictional damping and infinite memory controls all acting on the transversal displacement. We show that the system is well-posed in the sens of semigroup and that, under appropriate assumptions on the weights of the delay and the history data, the stability of the system holds in case of the equal-speed propagation as well as in the opposite case in spite of the presence of a discrete time delay, where the decay rate of solutions is given in terms of the smoothness of the initial data and the growth of the relaxation kernel at infinity. The results of this paper extend the ones obtained by the present author and Messaoudi in (Acta Math Sci 36:1–33, 2016) to the case of presence of discrete delay
Analysis of some Gastronomic Culturemes in the Extremely Contemporary Beur Novel Symbolic Images and Evoked Stories
No full textWhat subtle links between gastronomic culturemes and matrimonial customs can be revealed in the course of an extremely contemporary novel that invites us to taste the flavors of culinary art and those of an art of writing? The ineffable and unexpected answer is undoubtedly to be found in the refined intuition of an author who knows how to combine and cleverly merge community traditions of origin and Westernity in a test of the fleeting signs of interculturality. However, at the end of the road, the characters rebel, eager for freedom, still undecided between the desire to become Westernized and the need to preserve their Maghrebines
Use of modern GPUs in Design Optimisation
No full textGraphics Processing Units (GPUs) are a promising alternative hardware to CentralProcessing Units (CPU) for accelerating applications with a high computational power demand. In many fields researchers are taking advantage of the high computational power present in GPUs to speed up their applications. These applications span from data mining to machine learning and life sciences. The field of design optimization in particular benefits from this alternative hardware. The automated search on the design space has been delegated to GPUs or to a system of CPUs assisted by GPUs. This paper is among the firsts to review the use of GPUs especially for design optimization. The focus is on topology optimization, shape optimization and multidisciplinary design optimization (MDO). The target is to provide an overview not only on the progress madein design optimization using GPUs but also to highlights limitations that researchers have to cope with and the areas that require more research.Numerical Analysi
Women and politics in Algeria: essays on political representation
No full textGender and politics scholars have sought to determine whether there is a link between women’s descriptive representation, operationalized as the proportion of seats held by women in a national legislature, and women’s substantive representation, usually operationalized as laws that advance women’s rights. But, except for a few studies, the Arab world has not received significant attention which is surprising because the region has experienced a significant increase in women’s presence in politics. Further, most of this work has focused on democratic contexts, obscuring whether hypothesized links between women’s descriptive and substantive representation work in the same way in authoritarian contexts. To fill these gaps in the literature, I focus on the case of Algeria, where women’s presence in parliament increased from 8% to 31.6% after the adoption of a gender quota in 2012.
Drawing on in-depth interviews with a wide range of stakeholders, my research examines the backgrounds of women elected, constituency service priorities, legislative dynamics, and women’s agency. I argue that the “authoritarian toolkit,” i.e., the resources available to authoritarian governments to manage and control political outcomes, shapes women’s descriptive, substantive, and symbolic representation in ways that are distinct from how these dynamics operate in more democratic contexts.
While women parliamentarians reject the notion that they have an obligation to introduce and pass women’s rights laws, they invest time in helping their male and female constituents solve their everyday problems to challenge the notion that women do not belong in politics. Therefore, there may not be strong links between women's descriptive and substantive representation. However, the efforts of elected women on behalf of their male constituents may advance women’s symbolic representation by demonstrating women's abilities in the political realm.Ph.D.Includes bibliographical reference
Semi-supervised Statistical Approach for Network Anomaly Detection
No full textAbstractIntrusion Detection Systems (IDS) have become a very important defense measure against security threats. In recent years, computer networks are widely deployed for critical and complex systems, which make them more vulnerable to network attacks. In this paper, we propose a two-stage Semi-supervised Statistical approach for Anomaly Detection (SSAD). The first stage of SSAD aims to build a probabilistic model of normal instances and measures any deviation that exceeds an established threshold. This threshold is deduced from a regularized discriminant function of Maximum Likelihood (ML). The purpose of the second stage is to reduce False Alarm Rate (FAR) through an iterative process that reclassifies anomaly cluster, from the first stage, using a similarity distance and anomaly's cluster dispersion rate. We evaluate the proposed approach on the well-known intrusion detection dataset NSL-KDD and Kyoto 2006+. The experimental results show that SSAD outperforms the Naïve Bayes methods in terms of Detection Rate and False Positive Rate
- …
