The shear capacity of reinforced concrete pier caps in existing bridge support systems can be a factor which limits the capacity of an existing bridge. In their usual configuration, pier caps behave as deep beams and have the ability to carry load through tied arch action after the formation of diagonal cracks. Externally bonded fiber reinforced polymer (FRP) reinforcement has been shown to increase the shear capacity of reinforced concrete members which carry load through beam action. However, there is an insufficient amount of research to make it a viable strengthening system for beams which carry load through arch action, such as pier caps. Accordingly, this research was aimed at investigating the behavior of reinforced concrete pier caps through a coordinated experimental and analytical program and to recommend an external strengthening method for pier caps with perceived deficiencies in shear strength. The experimental study was performed on laboratory specimens based on an existing bridge in Georgia. A number of factors were examined, including size, percentage longitudinal reinforcement and crack control reinforcement. The results showed that increasing the longitudinal tension reinforcement increased the beam capacity by changing the shape of the tied arch. In contrast, the presence of crack control reinforcement did not change the point at which diagonal cracking occurred, but it did increase the ultimate capacity by reinforcing the concrete against splitting. The results of the experimental study were used in conjunction with a larger database to examine different analytical methods for estimating the ultimate capacity of deep beams, and a new method was developed for the design of external strengthening. Two specimens were tested with externally bonded FRP reinforcement applied longitudinally to increase the strength of the tension tie. The test results correlated well with the proposed method of analysis and showed that increasing the strength of the longitudinal tension tie is an effective way to increase the strength of a reinforced concrete deep beam.

<div>Autonomous transportation is a research field that has gained huge interest in recent years, with autonomous electric or hydrogen cars coming ever closer to seeing everyday use. Not just cars are subject to autonomous research though, the field of aviation is also being explored for fully autonomous flight. One very important aspect for making autonomous flight a reality is attitude control, the control of roll, pitch, and sometimes yaw. Traditional approaches for automated attitude control use PID (proportional-integral-derivative) controllers, which use hand-tuned parameters to fulfill the task. In this work, however, the use of Reinforcement Learning algorithms for attitude control will be explored. With the surge of more and more powerful artificial neural networks, which have proven to be universally usable function approximators, Deep Reinforcement Learning also becomes an intriguing option. </div><div>A software toolkit will be developed and used to allow for the use of multiple flight simulators to train agents with Reinforcement Learning as well as Deep Reinforcement Learning. Experiments will be run using different hyperparamters, algorithms, state representations, and reward functions to explore possible options for autonomous attitude control using Reinforcement Learning.</div>

Convolutional Neural Networks (CNNs) have been proved to deliver great results in the area of computer vision, however, one fundamental bottleneck with CNNs is the fact that it is heavily dependant on the ground truth, that is, labeled training data. A labeled dataset is a group of samples that have been tagged with one or more labels. In this degree project, we mitigate the data greedy behavior of CNNs by applying deep active learning with various kinds of sampling strategies. The main focus will be on the sampling strategies random sampling, least confidence sampling, margin sampling, entropy sampling, and K- means sampling. We choose to study the random sampling strategy since it will work as a baseline to the other sampling strategies. Moreover, the least confidence sampling, margin sampling, and entropy sampling strategies are uncertainty based sampling strategies, hence, it is interesting to study how they perform in comparison with the geometrical based K- means sampling strategy. These sampling strategies will help to find the most informative/representative samples amongst all unlabeled samples, thus, allowing us to label fewer samples. Furthermore, the benchmark datasets MNIST and CIFAR10 will be used to verify the performance of the various sampling strategies. The performance will be measured in terms of accuracy and less data needed. Lastly, we concluded that by using least confidence sampling and margin sampling we reduced the number of labeled samples by 79.25% in comparison with the random sampling strategy for the MNIST dataset. Moreover, by using entropy sampling we reduced the number of labeled samples by 67.92% for the CIFAR10 dataset. / Faltningsnätverk har visat sig leverera bra resultat inom området datorseende, men en fundamental flaskhals med Faltningsnätverk är det faktum att den är starkt beroende av klassificerade datapunkter. I det här examensarbetet hanterar vi Faltningsnätverkens giriga beteende av klassificerade datapunkter genom att använda deep active learning med olika typer av urvalsstrategier. Huvudfokus kommer ligga på urvalsstrategierna slumpmässigt urval, minst tillförlitlig urval, marginal baserad urval, entropi baserad urval och K- means urval. Vi väljer att studera den slumpmässiga urvalsstrategin eftersom att den kommer användas för att mäta prestandan hos de andra urvalsstrategierna. Dessutom valde vi urvalsstrategierna minst tillförlitlig urval, marginal baserad urval, entropi baserad urval eftersom att dessa är osäkerhetsbaserade strategier som är intressanta att jämföra med den geometribaserade strategin K- means. Dessa urvalsstrategier hjälper till att hitta de mest informativa/representativa datapunkter bland alla oklassificerade datapunkter, vilket gör att vi behöver klassificera färre datapunkter. Vidare kommer standard dastaseten MNIST och CIFAR10 att användas för att verifiera prestandan för de olika urvalsstrategierna. Slutligen drog vi slutsatsen att genom att använda minst tillförlitlig urval och marginal baserad urval minskade vi mängden klassificerade datapunkter med 79, 25%, i jämförelse med den slumpmässiga urvalsstrategin, för MNIST- datasetet. Dessutom minskade vi mängden klassificerade datapunkter med 67, 92% med hjälp av entropi baserad urval för CIFAR10datasetet.

This research is aimed at examining experimentally and analytically the behavior of reinforced concrete bridge pier caps strengthened with externally bonded reinforcement. In the experimental study, nine full-scale reinforced concrete bridge pier caps were built, externally strengthened with stainless steel reinforcement, and ten tested to failure. Load, deflection, and strain measurements were collected and two potential failure mechanisms were identified. In the analytical part of this work, mechanics-based equations were developed for calculating the shear strength of these types of structural elements when a diagonal shear crack is formed under loading. In addition, a combined strut-and-tie/truss model is proposed for determining the strength of reinforced concrete bridge caps with externally bonded reinforcement. Results from both experimental and analytical studies were compared and design recommendations are made for future adoption in bridge and building codes and specifications.

The South African gold mining sector consumes 47% of the mining industry’s electricity. On a deep level gold mine, 20% of the energy is consumed by the refrigeration system. The refrigeration system cools 67 ˚C virgin rock temperatures underground. Underground cooling demand increases significantly with deeper mining activities. Various cooling systems are available for underground cooling. This study focuses on the electricity usage of an ice storage system versus a chilled water system for underground cooling. An energy-savings approach was developed to determine possible power savings on the surface refrigeration system of Mine M. The savings approach involved converting an ice storage system to a chilled water system and varying the water flow through the system. The water flow was varied by installing variable speed drives on the evaporator and condenser water pumps. The feasibility of the energy-efficiency approach was simulated with a verified simulation model. Simulation results indicated the feasibility of converting the thermal ice storage to a chilled water system and implementing the energy-efficiency approach on Mine M. Simulated results indicated a 9% electricity saving when using a chilled water system. Various problems encountered by the mine were also a motivation to convert the thermal ice storage system. Converting an ice storage facility to a chilled water system for energy efficiency on a deep level gold mine Energy management is achieved through the monitoring, controlling and reporting of the implemented savings approach. Converting the glycol plant and recommissioning the chilled water plant gave the mine an additional chiller as backup to sufficiently meet underground demand. An annual summer power saving of 1.5 MW was achieved through the conversion and control strategy. It is concluded that conversion of the thermal ice storage system on Mine M results in an energy- and cost saving. / MIng (Mechanical Engineering), North-West University, Potchefstroom Campus, 2015

The South African gold mining sector consumes 47% of the mining industry’s electricity. On a deep level gold mine, 20% of the energy is consumed by the refrigeration system. The refrigeration system cools 67 ˚C virgin rock temperatures underground. Underground cooling demand increases significantly with deeper mining activities. Various cooling systems are available for underground cooling. This study focuses on the electricity usage of an ice storage system versus a chilled water system for underground cooling. An energy-savings approach was developed to determine possible power savings on the surface refrigeration system of Mine M. The savings approach involved converting an ice storage system to a chilled water system and varying the water flow through the system. The water flow was varied by installing variable speed drives on the evaporator and condenser water pumps. The feasibility of the energy-efficiency approach was simulated with a verified simulation model. Simulation results indicated the feasibility of converting the thermal ice storage to a chilled water system and implementing the energy-efficiency approach on Mine M. Simulated results indicated a 9% electricity saving when using a chilled water system. Various problems encountered by the mine were also a motivation to convert the thermal ice storage system. Converting an ice storage facility to a chilled water system for energy efficiency on a deep level gold mine Energy management is achieved through the monitoring, controlling and reporting of the implemented savings approach. Converting the glycol plant and recommissioning the chilled water plant gave the mine an additional chiller as backup to sufficiently meet underground demand. An annual summer power saving of 1.5 MW was achieved through the conversion and control strategy. It is concluded that conversion of the thermal ice storage system on Mine M results in an energy- and cost saving. / MIng (Mechanical Engineering), North-West University, Potchefstroom Campus, 2015

Master of Science / Department of Architectural Engineering and Construction Science / Kimberly W. Kramer / Strut-and-Tie models are useful in designing reinforced concrete structures with discontinuity regions where linear stress distribution is not valid. Deep beams are typically short girders with a large point load or multiple point loads. These point loads, in conjunction with the depth and length of the members, contribute to a member with primarily discontinuity regions. ACI 318-08 Building Code Requirements for Structural Concrete provides a method for designing deep beams using either Strut-and-Tie models (STM) or Deep Beam Method (DBM). This report compares dimension requirements, concrete quantities, steel quantities, and constructability of the two methods through the design of three different deep beams. The three designs consider the same single span deep beam with varying height and loading patterns. The first design is a single span deep beam with a large point load at the center girder. The second design is the deep beam with the same large point load at a quarter point of the girder. The last design is the deep beam with half the load at the midpoint and the other half at the quarter point. These three designs allow consideration of different shear and STM model geometry and design considerations. Comparing the two different designs shows the shear or cracking control reinforcement reduces by an average 13% because the STM considers the extra shear capacity through arching action. The tension steel used for either flexure or the tension tie increases by an average of 16% from deep beam in STM design. This is due to STM taking shear force through tension in the tension reinforcement through arching action. The main advantage of the STM is the ability to decreased member depth without decreasing shear reinforcement spacing. If the member depth is not a concern in the design, the preferred method is DBM unless the designer is familiar with STMs due to the similarity of deep beam and regular beam design theory.

Static authentication, where user identity is checked once at login time, can be circumvented no matter how strong the authentication mechanism is. Through attacks such as man-in-the-middle and its variants, an authenticated session can be hijacked later after the initial login process has been completed. In the last decade, continuous authentication (CA) using biometrics has emerged as a possible remedy against session hijacking. CA consists of testing the authenticity of the user repeatedly throughout the authenticated session as data becomes available. CA is expected to be carried out unobtrusively, due to its repetitive nature, which means that the authentication information must be collectible without any active involvement of the user and without using any special purpose hardware devices (e.g. biometric readers). Stylometry analysis, which consists of checking whether a target document was written or not by a specific individual, could potentially be used for CA. Although stylometric techniques can achieve high accuracy rates for long documents, it is still challenging to identify an author for short documents, in particular when dealing with large author populations. In this dissertation, we propose a new framework for continuous authentication using authorship verification based on the writing style. Authorship verification can be checked using stylometric techniques through the analysis of linguistic styles and writing characteristics of the authors. Different from traditional authorship verification that focuses on long texts, we tackle the use of short messages. Shorter authentication delay (i.e. smaller data sample) is essential to reduce the window size of the re-authentication period in CA. We validate our method using different block sizes, including 140, 280, and 500 characters, and investigate shallow and deep learning architectures for machine learning classification. Experimental evaluation of the proposed authorship verification approach based on the Enron emails dataset with 76 authors yields an Equal Error Rate (EER) of 8.21% and Twitter dataset with 100 authors yields an EER of 10.08%. The evaluation of the approach using relatively smaller forgery samples with 10 authors yields an EER of 5.48%. / Graduate

Malware continues to be one of the primary tools employed by attackers. It is used in attacks ranging from click fraud to nation state espionage. Malware infects hosts over the network through drive-by downloads and social engineering. These infected hosts communicate with remote command and control (C&C) servers to perform tasks and exfiltrate data. Malware's reliance on the network provides an opportunity for the detection and annotation of malicious communication. This thesis presents four main contributions. First, we design and implement a novel incident investigation system, named WebWitness. It automatically traces back and labels the sequence of events (e.g., visited web pages) preceding malware downloads to highlight how users reach attack pages on the web; providing a better understanding of current attack trends and aiding in the development of more effective defenses. Second, we conduct the first systematic study of modern web based social engineering malware download attacks. From this study we develop a categorization system for classifying social engineering downloads and use it to measure attack properties. From these measurements we show that it is possible to detect the majority of social engineering downloads using features from the download path. Third, we design and implement ExecScent, a novel system for mining new malware C&C domains from live networks. ExecScent automatically learns C&C traffic models that can adapt to the deployment network's traffic. This adaptive approach allows us to greatly reduce the false positives while maintaining a high number of true positives. Lastly, we develop a new packet scheduling algorithm for deep packet inspection that maximizes throughput by optimizing for cache affinity. By scheduling for cache affinity, we are able to deploy our systems on multi-gigabit networks.

Our understanding of the pathophysiology Parkinson’s disease has transformed over the last decade as we have come to appreciate the importance of changes in neuronal firing pattern that occur within the motor network in the dopamine deficient state. These changes in firing pattern, particularly increased synchrony result in oscillations that can be recorded as local field potentials. This thesis concerns itself with the study of beta oscillations which are characteristic of Parkinson’s disease. Firstly, I investigate whether beta oscillations play a pathophysiological role in Parkinson’s disease or whether they are purely epiphenomenal by augmenting beta with low frequency deep brain stimulation. In this study I show that rigidity is increased by ~25% with low frequency stimulation providing significant further evidence for a patho-physiological role of beta in Parkinson’s disease. Next I investigate whether beta oscillations correlate with Parkinsonian severity at rest and could therefore potentially be used as a biomarker of clinical state. I demonstrate that the variability of beta amplitude recorded from the subthalamic nucleus strongly correlates with symptom severity at rest and also in response to levodopa administration. I then use beta amplitude as a biomarker for a trial of adaptive deep brain stimulation in Parkinson’s disease. I show that by using beta amplitude to control stimulation, time on stimulation is reduced by >50% but despite this, clinical outcome is improved by 25% relative to conventional continuous high frequency stimulation. Finally, I investigate the bilateral subcortical beta network and its response to levodopa. I report statistically significant bilateral functional connectivity in the beta range which is driven by phase locking and modulated by levodopa in the low beta range with implications for bilateral adaptive deep brain stimulation. These findings further our understanding of the pathophysiological role of beta oscillations in Parkinson’s disease and provide new avenues for treatment development.