Orientador: Emilson Pereira Leite / Dissertação (mestrado) - Universidade Estadual de Campinas, Instituto de Geociências / Made available in DSpace on 2018-08-27T16:08:56Z (GMT). No. of bitstreams: 1 Mori_RobertaTomi_M.pdf: 3320202 bytes, checksum: 0cc28a2902a082cdb74ee3af69f82776 (MD5) Previous issue date: 2015 / Resumo: Uma boa caracterização geológica dos reservatórios é de grande importância para a diminuição dos riscos da perfuração de um poço seco, assim como os custos de exploração e desenvolvimento de tal reservatório. No presente trabalho, buscou-se predizer os valores de porosidade de um reservatório através da integração de dados sísmicos 3D com perfis geofísicos de poços através de dois métodos: Regressão Linear Multiatributo e Redes Neurais Artificiais. O reservatório em questão é de constituição carbonática de origem Albiana, do final do Cretáceo Inferior. Na primeira etapa, foram traçados horizontes baseando-se em eventos sísmicos contínuos nas seções sísmicas, chamados superfícies cronoestratigráficas, no intervalo de profundidade que abrange o reservatório. Na segunda etapa, foram obtidos predições dos valores de porosidade do reservatório, assim como os respectivos modelos 3D, através dos dois métodos acima citados. Com a RLM, foram obtidos valores altos, baixos e intermediários de porosidade, variando desde 5% até 40%. Já com a RNA, tais valores variaram de 5% a 30%. Em ambos os métodos, os valores de porosidade apresentaram um crescimento da porção sudoeste em direção à porção nordeste, apresentando baixos valores nas profundidades referentes aos horizontes traçados. Picos com os valores máximos de porosidade foram observados em pontos dispersos por todo o volume 3D. Comparando-se os resultados obtidos pelos dois métodos, ficou evidente a maior eficácia da RNA, a qual apresentou correlação de 0,90 entre os valores reais e os valores preditos e erro de 2,86%, enquanto que os resultados obtidos através da RLM apresentou correlação de 0,55 e erro de 5,45%. Além disso, foi feita uma comparação com os aspectos geológicos do reservatório, na qual concluiu-se que as baixas porosidades da porção sudoeste se deve à presença de microporosidade e as altas porosidades da porção nordeste, à macroporosidade original das rochas. Concluiu-se também que as baixas porosidades encontradas nas regiões dos horizontes sísmicos estão relacionadas às diferentes texturas de rochas, já que as rochas presentes nessas regiões possuem maiores quantidades de matriz carbonática (packstones e wackestones) quando comparadas com as rochas das regiões entre os horizontes (grainstones) / Abstract: A good geological reservoirs characterization is very important for reducing the risk of drilling a dry hole as well as the costs for reservoir exploration and development. In this study, it was attempted to predict the porosity values of a reservoir through the integrations of 3D seismic data with geophysical well logs using two different methods: Multiattribute Linear Regression and Artificial Neural Networks. The studied reservoir has a carbonate composite, with the age of Albian, in late Early Cretaceous. On the first stage of the study, horizons have been traced based on continuous seismic events on seismic sections, in depths that cover the reservoir. On the second stage, it was obtained some predictions of reservoir porosity values, as well as their 3D models by the two methods that was already mentioned. High, low and intermediate porosity values have been obtained by the MLR, ranging from 5% to 40%. With the ANN, these values ranged from 5% to 30%. In both methods, the porosity values grew from south-west portion toward the northeast portion, with low values on the depths related to the horizons traced. We can observe maximum value peaks of porosity at points scattered throughout the 3D volume. A comparison of the results obtained by the two methods evidence the greater efficiency of the ANN, with a correlation of 0,90, between actual porosity and predicted values, and 2.86% of error, while the results obtained by the MLR showed a correlation of 0,55 and an error of 5.45%. Furthermore, we have made a comparison between the results obtained and the reservoir geological features, which allows us to conclude that the low porosity in the south-west portion is because of microporosity, while the high porosity in the northeast is because of the original macroporosity of the rocks. We also conclude that low porosity found on horizon surfaces are related to different rock textures, once the rocks on these horizon regions have more carbonatic matrix in their constitution (packstones and wackestones) than the rocks in the other regions between the horizons (grainstones) / Mestrado / Geologia e Recursos Naturais / Mestra em Geociências

Les architectures d’ordinateur classiques sont optimisées pour le traitement déterministe d’informations pré-formatées et ont donc des difficultés avec des données naturelles bruitées (images, sons, etc.). Comme celles-ci deviennent nombreuses, de nouveaux circuits neuromorphiques (inspirés par le cerveau) tels que les réseaux de neurones émergent. Des nano-dispositifs, appelés memristors, pourraient permettre leur implémentation sur puce avec une haute efficacité énergétique et en s’approchant de la haute connectivité synaptique du cerveau.Dans ce travail, nous étudions des memristors basés sur des jonctions tunnel ferroélectriques qui sont composées d’une couche ferroélectrique ultramince entre deux électrodes métalliques. Nous montrons que le renversement de la polarisation de BiFeO3 induit des changements de résistance de quatre ordres de grandeurs et établissons un lien direct entre les états de domaines mixtes et les niveaux de résistance intermédiaires.En alternant les matériaux des électrodes, nous révélons leur influence sur la barrière électrostatique et les propriétés dynamiques des memristors. Des expériences d’impulsion unique de tension montrent un retournement de polarisation ultra-rapide. Nous approfondissons l’étude de cette dynamique par des mesures d’impulsions cumulées. La combinaison de leur analyse avec de l’imagerie par microscopie à force piézoélectrique nous permet d’établir un modèle dynamique du memristor. Suite à la démonstration de la spike-timing-dependent plasticity, une règle d’apprentissage importante, nous pouvons prédire le comportement de notre synapse artificielle. Ceci représente une avance majeure vers la réalisation de réseaux de neurones sur puce dotés d’un auto-apprentissage non-supervisé. / Classical computer architectures are optimized to process pre-formatted information in a deterministic way and therefore struggle to treat unorganized natural data (images, sounds, etc.). As these become more and more important, the brain inspires new, neuromorphic computer circuits such as neural networks. Their energy-efficient hardware implementations will greatly benefit from nanodevices, called memristors, whose small size could enable the high synaptic connectivity degree observed in the brain.In this work, we concentrate on memristors based on ferroelectric tunnel junctions that are composed of an ultrathin ferroelectric film between two metallic electrodes. We show that the polarization reversal in BiFeO3 films can induce resistance contrasts as high as 10^4 and how mixed domain states are connected to intermediate resistance levels.Changing the electrode materials provides insights into their influence on the electrostatic barrier and dynamic properties of these memristors. Single-shot switching experiments reveal very fast polarization switching which we further investigate in cumulative measurements. Their analysis in combination with piezoresponse force microscopy finally allows us to establish a model describing the memristor dynamics under arbitrary voltage signals. After the demonstration of an important learning rule for neural networks, called spike-timing-dependent plasticity, we successfully predict new, previously unexplored learning curves. This constitutes an important step towards the realization of unsupervised self-learning hardware neural networks.

Artificial intelligence has become a hot topic in the past couple of years because of its potential of solving problems. The most used subset of artificial intelligence today is machine learning, which is essentially the way a machine can learn to do tasks without getting any explicit instructions. A problem that has historically been solved by common knowledge and experience is the planning of bus transportation, which has been prone to mistakes. This thesis investigates how to extract the key features of a raw dataset and if a couple of machine learning algorithms can be applied to predict and plan the public bus transportation, while also considering the weather conditions. By using a pre-processing method to extract the features before creating and evaluating an k-nearest neighbors model as well as an artificial neural network model, predicting the passenger count on a given route could help planning of the bus transportation. The outcome of the thesis was that the feature extraction was successful, and both models could successfully predict the passenger count based on normal conditions. However, in extreme conditions such as the pandemic during 2020, the models could not be proven to successfully predict the passenger count nor being used to plan the bus transportation.

<p>Due to the rapid development of wireless communication and network technology, more and more wireless devices (e.g., Siemens, Lutron, etc.) are used in residential and commercial buildings. The wireless system has many advantages that traditional wired-based systems do not have, such as time-saving deployment and easy maintenance. However, the wireless system is also vulnerable to cyber-attacks since the data packets are transmitted by radio waves rather than by physical medium. The current cyber detection system (e.g., Intrusion detection system) monitors the data traffic to identify the anomalies in the network. However, it is unable to detect the attacks that tamper with the control logic or operating parameters, which results in the malfunction of the system. This thesis developed an integrated, cyber-security framework for cyber-attack detection in smart buildings.</p> <p>The objective of this research is to develop an integrated cyber-security framework for wireless-based smart building systems to protect buildings from the cyber-attacks. The wireless-based smart building systems are operated and controlled by either a two-position or continuous controlled approach. The efforts in this study have developed a cyber-security framework to deal with both two-position control and continuous control. For the two-position controlled smart buildings, the developed cyber-security framework integrates the data and models of both cyber and physical domains of building systems to detect faults, abnormal operations, and cyber attacks. The cyber-security framework developed for the continuous controlled system combines a data-driven model for detecting the faults of sensor measurements and a physical model based on engineering principle (e.g., laws of thermodynamics or control logic) to detect the anomaly of system operation.</p> <p>To develop the cyber-security frameworks, the testbeds corresponding to the two-position and continuous wireless systems were constructed for attack-oriented tests. A wireless-based lighting system for smart homes was used as the testbed for the study of the two-position control. It has a wireless occupancy sensor, an actuator for the light switch, and an open-source operating platform (OpenHAB) for system control and monitor. The platform of the wireless is the ZigBee. An indoor shading system at a living lab in new Herrick building at Purdue University was utilized as the testbed for the study of the continuous controlled system. The indoor shading system exploits the roller shades to block the excess daylighting to provide an acceptable illuminance condition for occupants. The shading system uses the wireless illuminance sensor, weather condition, and wire-based controller to automatically operate the shades for the acceptable illuminance. </p> <p>The study implemented designed cyber-attacks to validate the effectiveness of the developed frameworks. The final results show that the developed two models were able to detect the attacks effectively (95-100% attacks identified and isolated). The abnormal operations tested in two-position control system were identified when an abnormal state was triggered, or the modelled state and real state did not match in the finite state machine model developed. For continuous control, the abnormal operations were detected when there is a significant deviation between the modelled measurement and the actual measurement. The cybersecurity framework developed in the thesis demonstrates an effective approach for detecting system faults caused by attacks. The frameworks could be widely used for other different building systems and beyond buildings, such as transportation or industrial manufacturing systems.</p>

abstract: In order to deploy autonomous multi-robot teams for humanitarian demining in Colombia, two key problems need to be addressed. First, a robotic controller with limited power that can completely cover a dynamic search area is needed. Second, the Colombian National Army (COLAR) needs to increase its science, technology and innovation (STI) capacity to help develop, build and maintain such robots. Using Thangavelautham's (2012, 2017) Artificial Neural Tissue (ANT) control algorithm, a robotic controller for an autonomous multi-robot team was developed. Trained by a simple genetic algorithm, ANT is an artificial neural network (ANN) controller with a sparse, coarse coding network architecture and adaptive activation functions. Starting from the exterior of open, basic geometric grid areas, computer simulations of an ANT multi-robot team with limited time steps, no central controller and limited a priori information, covered some areas completely in linear time, and other areas near completely in quasi-linear time, comparable to the theoretical cover time bounds of grid-based, ant pheromone, area coverage algorithms. To mitigate catastrophic forgetting, a new learning method for ANT, Lifelong Adaptive Neuronal Learning (LANL) was developed, where neural network weight parameters for a specific coverage task were frozen, and only the activation function and output behavior parameters were re-trained for a new coverage task. The performance of the LANL controllers were comparable to training all parameters ab initio, for a new ANT controller for the new coverage task. To increase COLAR's STI capacity, a proposal for a new STI officer corps, Project ÉLITE (Equipo de Líderes en Investigación y Tecnología del Ejército) was developed, where officers enroll in a research intensive, master of science program in applied mathematics or physics in Colombia, and conduct research in the US during their final year. ÉLITE is inspired by the Israel Defense Forces Talpiot program. / Dissertation/Thesis / Doctoral Dissertation Applied Mathematics for the Life and Social Sciences 2019

Master thesis deals with possibilities of playing games using artificial intelligence. Also, there is presented artificial intelligence with ability of playing selected game. Artificial intelligence presented in this work is considered as artificial neural networks. First theoretical part is focused on description worldwide successful and known artificial intelligence strategies. After that there is brief description of neural networks and analysis of some libraries for machine learning. Practical part is focused on implementation of created game with Python programming language and in the end, there is analysis of designed solution.

Prediction of both big and small decisions is something most companies have to make on a daily basis. The importance of having a highly accurate technique for different decision-making is not something that is new. However, even though the importance of prediction is a fact to most people, current techniques for estimation are still often highly inaccurate. The consequences of an inaccurate prediction can be huge in the differences between the misclassifications. Not just in the industry but for many different areas. Machine learning have in the recent couple of years improved significantly and are now considered a reliable method to use for prediction. The main goal of this research is to predict casting defects with the help of a machine-learning algorithm based on process parameters. In order to achieve the main goal, some sub-objectives have been identified to successfully reach those goals. A problem when dealing with machine learning is an unbalanced dataset. When training a network, it is essential that the dataset is balanced. In this research we have successfully balanced the dataset. Undersampling was the method used in our research to establish our balanced dataset. The research compares and evaluates a couple of different undersample methods in order to see which undersampling is best suited for this project. Three different machine models, “random forest”, “artificial neural network”, and “k-nearest neighbor”, are also compared to each other to see what model performs best. The conlcusion reached was that the best method for both undersampling and machine learning model varied due to many different reasons. So, in order to find the best model with the best method for a specific job, all the models and methods need to be tested. However, the undersampling method that provided best performances most times in our research was the NearMiss version 2 model. Artificial Neural Network was the machine learning model that had most success in our research. It performed best in two out of three evaluations and comparisons.

This master‘s thesis deals with the use of higher order cumulants for classification of cardiac cycles. Second-, third-, and fourth-order cumulants were calculated from ECG recorded in isolated rabbit hearts during experiments with repeated ischemia. Cumulants properties useful for the subsequent classification were verified on ECG segments from control and ischemic group. The results were statistically analyzed. Cumulants are then used as feature vectors for classification of ECG segments by means of artificial neural network.

This diploma thesis is dealing with the impact of oxide scales on heat conduction. One of the main tools that were used are numerical simulations. Heat conduction is modelled by solving partial differential equations. Regression models and artificial neural networks are used for the prediction of the influence of oxides on cooling intensity. Determination of the conditions when the cooling was intensified and comparison of individual methods of prediction are the main results of the thesis.

Widespread use of wireless networks has made security a serious issue. This thesis proposes misuse based intrusion detection system for wireless networks, which applies artificial neural network to captured frames for purpose of anomalous patterns recognition. To address the problem of high positive alarm rate, this thesis presents a method of applying two artificial neural networks.