Previsão da Classificação da qualidade dos efluentes da indústria petrolífera com recurso a sistemas inteligentes

O presente trabalho realizou-se na Refinaria de Sines e teve como principal objectivo a utilização de ferramentas oriundas da Área Científica da Inteligência Artificial no desenvolvimento de modelos de previsão da classificação da Água Residual Industrial de acordo com a Legislação em vigor, com vista à minimização dos impactes ambientais e das tarifas aplicadas pela Concessionária (Águas de Santo André) à Refinaria. Actualmente a avaliação da qualidade do efluente é realizada através de métodos analíticos após colheita de uma amostra do efluente final. Esta abordagem é muito restritiva já que não permite actuar sobre o efluente em questão pois apenas pode evitar que, no futuro, uma mistura semelhante volte a ser refinada. Devido a estas limitações, o desenvolvimento de modelos de previsão baseados em Data Mining mostrou ser uma alternativa para uma questão pró-activa da qualidade dos efluentes que pode contribuir decisivamente para o cumprimento das metas definidas pela Empresa. No decurso do trabalho, foram desenvolvidos dois modelos de previsão da qualidade do efluente industrial com desempenhos muito semelhantes. Um deles utiliza a composição das misturas processadas e o outro, utiliza informações relativas ao crude predominante na mistura. ABSTRACT; This study has taken place at the Sines Refinery and its main objective is the use of Artificial Intelligence tools for the development of predictive models to classify industrial residual waters according with the Portuguese Law, based on the characteristics of the mixtures of crude oil that arrive into the Refinery to be processed, to minimize the Environmental impacts and the application of taxes. Currently, the evaluation of the quality of effluent is performed by analytical methods after harvesting a sample of the final effluent. This approach is very restrictive since it does not act on the intended effluent; it can only avoid that in the future a similar mixture is refined. Duet these limitations, the development of forecasting models based on Data Mining has proved to be an alternative on the important issue which is the quality of effluent, which may contribute to the achievement of targets set by the Company. During this study, two models were developed to predict the quality of industrial effluents with very similar performances. One uses the composition of processed mixtures and the other uses information regarding the predominant oil in the mixture.

Modelagem univariada e regressão múltipla na análise da oferta e do preço do boi gordo no mercado brasileiro

Dissertação (mestrado)—Universidade de Brasília, Faculdade de Agronomia e Medicina Veterinária, Programa de Pós-Graduação em Agronegócios, 2016.

Análise bioclimática do bairro do Renascença II São Luis-MA: realidade e perspectiva do conforto térmico em espaços externos

The present research concerns about outdoor s thermal comfort conditions in hot-humid climate cities, understanding that life quality is a result of the urban object s type built for the human being in an environment with specific climate and morphological characteristics. It is presented as object of study the correlation between the neighborhood Renascença II s microclimate in São Luis /MA-Brazil, hot-humid climate city, and its urban morphological changes. As well as the thermal comfort s satisfaction level of its outdoor users. The research has as general goal to diagnosis the way these transformations caused by the urbanization influence the Renascença II s microclimate, identifying critical spots of the studied area, in order to contribute with land use recommendations based on bioclimatic architecture concepts and supply bases to urban design decisions adequate to the São Luis climate. It is presented as theoretical bases the urban climate, its concepts and elements. After that, the thermal comfort conditioners and its prediction models of thermal comfort sensation in outdoor are presented. The predictive models are presented along with bioclimatic assessment methods. Finally the use of bioclimatic assessment as an effective tool to identify places that need changes or preservation in order to seek environment quality. The applied methodology was based on the studies of Katzschner (1997), complemented by Oliveira s (1988) and Bustos Romero s (2001) studies that suggest an analysis and evaluation of maps of topography, buildings floors, land use, green areas and land covering, in order to overlap their characteristics and identify climate variable s measurements points; then a quantitative analysis of the climate variables (air temperature and humidity, wind speed and direction) of the chosen points takes place. It was perceived that Renaissance II has no permanence areas as squares or parks, its outdoor has little vegetation and presets high land impermeability and built density levels. The majority of the people interviewed said that was comfortable in a range of air temperature between 27,28ºC and 30,71ºC. The elaboration of a neighborhood master plan is important, which defines strategies for improvement of the life quality of its inhabitants

Automatic learning of pre-miRNAs from different species.

Discovery of microRNAs (miRNAs) relies on predictive models for characteristic features from miRNA precursors (pre-miRNAs). The short length of miRNA genes and the lack of pronounced sequence features complicate this task. To accommodate the peculiarities of plant and animal miRNAs systems, tools for both systems have evolved differently. However, these tools are biased towards the species for which they were primarily developed and, consequently, their predictive performance on data sets from other species of the same kingdom might be lower. While these biases are intrinsic to the species, their characterization can lead to computational approaches capable of diminishing their negative effect on the accuracy of pre-miRNAs predictive models. We investigate in this study how 45 predictive models induced for data sets from 45 species, distributed in eight subphyla/classes, perform when applied to a species different from the species used in its induction. Results: Our computational experiments show that the separability of pre-miRNAs and pseudo pre-miRNAs instances is species-dependent and no feature set performs well for all species, even within the same subphylum/class. Mitigating this species dependency, we show that an ensemble of classifiers reduced the classification errors for all 45 species. As the ensemble members were obtained using meaningful, and yet computationally viable feature sets, the ensembles also have a lower computational cost than individual classifiers that rely on energy stability parameters, which are of prohibitive computational cost in large scale applications. Conclusion: In this study, the combination of multiple pre-miRNAs feature sets and multiple learning biases enhanced the predictive accuracy of pre-miRNAs classifiers of 45 species. This is certainly a promising approach to be incorporated in miRNA discovery tools towards more accurate and less species-dependent tools.

Evaluating Karner Blue Butterfly (Lycaeides Melissa Samuelis Nabokov) Habitat Selection in the State of Wisconsin, USA

The federally endangered Karner blue butterfly (Lycaeides melissa samuelis Nabokov) persists in rare oak/pine grassland communities spanning across the Great Lakes region, relying on host plant wild blue lupine (Lupinus perennis). Conservation efforts since 1992 have led to the development of several programs that restore and monitor habitat. This study aims to evaluate Karner blue habitat selection in the state of Wisconsin and develop high-resolution tools for use in conservation efforts. Spatial predictive models developed during this study accurately predicted potential habitat across state properties based on soils and canopy cover, and identified ~51-100% of Karner blue occurrences based on lupine and shrub/tree cover, and focal nectar plant abundance. When evaluated relative to American bison (Bison bison), Karner blues and lupine were more likely to occur in areas of low disturbance, but aggregated where bison were recently present in areas of moderate/high disturbance. Lupine C:N ratio increased relative to cover of shrubs/trees and focal nectar plant abundance and decreased relative to cover of groundlitter. Karner blue density increased with lupine C:N ratio, decreased with nitrogen content, and was not related to phenolic levels. We strongly suggest that areas of different soil textures must be managed differently and that maintenance techniques should generate a mix of shrubs/tree cover (10-45%), groundlitter cover (~10-40%), >5% cover of lupine, and establish an abundance of focal nectar plants. This study provides unique tools for use in conservation and should aid in focusing management efforts and recovery of this species.

GIS-integrated mathematical modeling of social phenomena at macro- and micro- levels—a multivariate geographically-weighted regression model for identifying locations vulnerable to hosting terrorist safe-houses: France as case study

Adaptability and invisibility are hallmarks of modern terrorism, and keeping pace with its dynamic nature presents a serious challenge for societies throughout the world. Innovations in computer science have incorporated applied mathematics to develop a wide array of predictive models to support the variety of approaches to counterterrorism. Predictive models are usually designed to forecast the location of attacks. Although this may protect individual structures or locations, it does not reduce the threat—it merely changes the target. While predictive models dedicated to events or social relationships receive much attention where the mathematical and social science communities intersect, models dedicated to terrorist locations such as safe-houses (rather than their targets or training sites) are rare and possibly nonexistent. At the time of this research, there were no publically available models designed to predict locations where violent extremists are likely to reside. This research uses France as a case study to present a complex systems model that incorporates multiple quantitative, qualitative and geospatial variables that differ in terms of scale, weight, and type. Though many of these variables are recognized by specialists in security studies, there remains controversy with respect to their relative importance, degree of interaction, and interdependence. Additionally, some of the variables proposed in this research are not generally recognized as drivers, yet they warrant examination based on their potential role within a complex system. This research tested multiple regression models and determined that geographically-weighted regression analysis produced the most accurate result to accommodate non-stationary coefficient behavior, demonstrating that geographic variables are critical to understanding and predicting the phenomenon of terrorism. This dissertation presents a flexible prototypical model that can be refined and applied to other regions to inform stakeholders such as policy-makers and law enforcement in their efforts to improve national security and enhance quality-of-life.

Predição da criatividade e saúde mental

A criatividade tem sido mencionada desde há muito tempo pela literatura. A forma de a conceptualizar tem dependido em muito dos investigadores e da sua visão sobre o que é criatividade, porém as suas potencialidades têm tomado cada vez mais interesse pela comunidade científica. Os seus benefícios estendem-se a diversas áreas da vida humana, como sentimentos de satisfação, bem-estar e saúde mental. A investigação aqui apresentada procura encontrar modelos preditivos de comportamentos criativos, pressupondo que estes poderão ser de grande auxílio para o desenvolvimento de um melhor bem-estar psicológico. Assim a investigação comportou o uso de quatro variáveis: a pessoa, o processo, o produto e o ambiente criativo. A amostra contou com 215 estudantes universitários e os resultados revelaram o encontro de quatro modelos preditivos de comportamentos criativos, com uma maior ênfase nas artes visuais e manuais. Conclui-se com a ideia de que a predição de produtos criativos é possível e que tal poderá ser preponderante para um bom desenvolvimento da saúde mental.

¬FISH DISTRIBUTIONS IN LAKE ONTARIO’S EASTERN BASIN AND THE UPPER ST. LAWRENCE RIVER: AN ANALYSIS USING GIS AND OCCUPANCY MODELLING TECHNIQUES

Predictive models of species distributions are important tools for fisheries management. Unfortunately, these predictive models can be difficult to perform on large waterbodies where fish are difficult to detect and exhaustive sampling is not possible. In recent years the development of Geographic Information Systems (GIS) and new occupancy modelling techniques has improved our ability to predict distributions across landscapes as well as account for imperfect detection. I surveyed the nearshore fish community at 105 sites between Kingston, Ontario and Rockport, Ontario with the objective of modelling geographic and environmental characteristics associated with littoral fish distributions. Occupancy modelling was performed on Round Goby, Yellow perch, and Lepomis spp. Modelling with geographic and environmental covariates revealed the effect of shoreline exposure on nearshore habitat characteristics and the occupancy of Round Goby. Yellow Perch, and Lepomis spp. occupancy was most strongly associated negatively with distance to a wetland. These results are consistent with past research on large lake systems indicate the importance of wetlands and shoreline exposure in determining the fish community of the littoral zone. By examining 3 species with varying rates of occupancy and detection, this study was also able to demonstrate the variable utility of occupancy modelling.

A next application prediction service using the BaranC framework

Predicting user behaviour enables user assistant services provide personalized services to the users. This requires a comprehensive user model that can be created by monitoring user interactions and activities. BaranC is a framework that performs user interface (UI) monitoring (and collects all associated context data), builds a user model, and supports services that make use of the user model. A prediction service, Next-App, is built to demonstrate the use of the framework and to evaluate the usefulness of such a prediction service. Next-App analyses a user's data, learns patterns, makes a model for a user, and finally predicts, based on the user model and current context, what application(s) the user is likely to want to use. The prediction is pro-active and dynamic, reflecting the current context, and is also dynamic in that it responds to changes in the user model, as might occur over time as a user's habits change. Initial evaluation of Next-App indicates a high-level of satisfaction with the service.

Predição da criatividade e saúde mental

A criatividade tem sido mencionada desde há muito tempo pela literatura. A forma de a conceptualizar tem dependido em muito dos investigadores e da sua visão sobre o que é criatividade, porém as suas potencialidades têm tomado cada vez mais interesse pela comunidade científica. Os seus benefícios estendem-se a diversas áreas da vida humana, como sentimentos de satisfação, bem-estar e saúde mental. A investigação aqui apresentada procura encontrar modelos preditivos de comportamentos criativos, pressupondo que estes poderão ser de grande auxílio para o desenvolvimento de um melhor bem-estar psicológico. Assim a investigação comportou o uso de quatro variáveis: a pessoa, o processo, o produto e o ambiente criativo. A amostra contou com 215 estudantes universitários e os resultados revelaram o encontro de quatro modelos preditivos de comportamentos criativos, com uma maior ênfase nas artes visuais e manuais. Conclui-se com a ideia de que a predição de produtos criativos é possível e que tal poderá ser preponderante para um bom desenvolvimento da saúde mental.

Load Profile Analysis Tool for Electrical Appliances in Households

This paper presents a methodology to forecast the hourly and daily consumption in households. The methodology was validated for households in Lisbon region, Portugal. The paper shows that the forecast tool allows obtaining satisfactory results for forecasting. Models of demand response allow the support of consumer’s decision in exchange for an economic benefit by the redefinition of load profile or changing the appliance consumption period. It is also in the interest of electric utilities to take advantage of these changes, particularly when consumers have an action on the demand-side management or production. Producers need to understand the load profile of households that are connected to a smart grid, to promote a better use of energy, as well as optimize the use of micro-generation from renewable sources, not only to delivering to the network but also in self-consumption.

Load Profile Analysis Tool for Electrical Appliances in Households Assisted by CPS

This paper presents a methodology to forecast the hourly and daily consumption in households assisted by cyber physical systems. The methodology was validated using a database of consumption of a set of 93 domestic consumers. Forecast tools used were based on Fast Fourier Series and Generalized Reduced Gradient. Both tools were tested and their forecast results were compared. The paper shows that both tools allow obtaining satisfactory results for energy consumption forecasting.

Uso da espectroscopia de reflectância do infravermelho próximo (NIRS) para previsão da composição bromatológica de vagens de algaroba e palma forrageira.

Espécies forrageiras adaptadas às condições semiáridas são uma alternativa para reduzir os impactos negativos na cadeia produtiva de ruminantes da região Nordeste brasileira devido à sazonalidade na oferta de forragem, além de reduzir custo com o fornecimento de alimentos concentrados. Dentre as espécies, a vagem de algaroba (Prosopis juliflora SW D.C.) e palma forrageira (Opuntia e Nopalea) ganham destaque por tolerarem o déficit hídrico e produzirem em períodos onde a oferta de forragem está reduzida, além de apresentam bom valor nutricional e serem bem aceitas pelos animais. Porém, devido à variação na sua composição, seu uso na alimentação animal exige o conhecimento profundo da sua composição para a elaboração de dietas balanceadas. No entanto, devido ao custo e tempo para análise, os produtores não fazem uso da prática de análise da composição químico-bromatológica dos alimentos. Por isto, a espectroscopia de reflectância no infravermelho próximo (NIRS) representa uma importante alternativa aos métodos tradicionais. Objetivou-se com este estudo desenvolver e validar modelos de predição da composição bromatológica de vagem de algaroba e palma forrageira baseados em espectroscopia NIRS, escaneadas em dois modelos de equipamentos e com diferentes processamentos da amostra. Foram coletadas amostras de vagem de algaroba nos estados do Ceará, Bahia, Paraíba e Pernambuco, e amostras de palma forrageira nos estados do Ceará, Paraíba e Pernambuco, frescas (in natura) ou pré-secas e moídas. Para obtenção dos espectros utilizaram-se dois equipamentos NIR, Perten DA 7250 e FOSS 5000. Inicialmente os alimentos foram escaneados in natura em aparelho do modelo Perten, e, com o auxílio do software The Unscrambler 10.2 foi selecionado um grupo de amostras para o banco de calibração. As amostras selecionadas foram secas e moídas, e escaneadas novamente em equipamentos Perten e FOSS. Os valores dos parâmetros de referência foram obtidos por meio de metodologias tradicionalmente aplicadas em laboratório de nutrição animal para matéria seca (MS), matéria mineral (MM), matéria orgânica (MO), proteína bruta (PB), estrato etéreo (EE), fibra solúvel em detergente neutro (FDN), fibra solúvel em detergente ácido (FDA), hemicelulose (HEM) e digestibilidade in vitro da matéria seca (DIVMS). O desempenho dos modelos foi avaliado de acordo com os erros médios de calibração (RMSEC) e validação (RMSECV), coeficiente de determinação (R2 ) e da relação de desempenho de desvio dos modelos (RPD). A análise exploratória dos dados, por meio de tratamentos espectrais e análise de componentes principais (PCA), demonstraram que os bancos de dados eram similares entre si, dando segurança de desenvolver os modelos com todas as amostras selecionadas em um único modelo para cada alimento, algaroba e palma. Na avaliação dos resultados de referência, observou-se que a variação dos resultados para cada parâmetro corroboraram com os descritos na literatura. No desempenho dos modelos, aqueles desenvolvidos com pré-processamento da amostra (pré-secagem e moagem) se mostraram mais robustos do que aqueles construídos com amostras in natura. O aparelho NIRS Perten apresentou desempenho semelhante ao equipamento FOSS, apesar desse último cobrir uma faixa espectral maior e com intervalos de leituras menores. A técnica NIR, associada ao método de calibração multivariada de regressão por meio de quadrados mínimos (PLS), mostrou-se confiável para prever a composição químico-bromatológica de vagem de algaroba e da palma forrageira. Abstract: Forage species adapted to semi-arid conditions are an alternative to reduce the negative impacts in the feed supply for ruminants in the Brazilian Northeast region, due to seasonality in forage availability, as well as in the reducing of cost by providing concentrated feedstuffs. Among the species, mesquite pods (Prosopis juliflora SW DC) and spineless cactus (Opuntia and Nopalea) are highlighted for tolerating the drought and producion in periods where the forage is scarce, and have high nutritional value and also are well accepted by the animals. However, its use in animal diets requires a knowledge about its composition to prepare balanced diets. However, farmers usually do not use feed composition analysis, because their high cost and time-consuming. Thus, the Near Infrared Reflectance Spectroscopy in the (NIRS) is an important alternative to traditional methods. The objective of this study to develop and validate predictive models of the chemical composition of mesquite pods and spineless cactus-based NIRS spectroscopy, scanned in two different spectrometers and sample processing. Mesquite pods samples were collected in the states of Ceará, Bahia, Paraiba and Pernambuco, and samples of forage cactus in the states of Ceará, Paraíba and Pernambuco. In order to obtain the spectra, it was used two NIR equipment: Perten DA 7250 and FOSS 5000. sSpectra of samples were initially obtained fresh (as received) using Perten instrument, and with The Unscrambler software 10.2, a group of subsamples was selected to model development, keeping out redundant ones. The selected samples were dried and ground, and scanned again in both Perten and FOSS instruments. The values of the reference analysis were obtained by methods traditionally applied in animal nutrition laboratory to dry matter (DM), mineral matter (MM), organic matter (OM), crude protein (CP), ether extract (EE), soluble neutral detergent fiber (NDF), soluble acid detergent fiber (ADF), hemicellulose ( HEM) and in vitro digestibility of dry matter (DIVDM). The performance of the models was evaluated according to the Root Mean Square Error of Calibration (RMSEC) and cross-validation (RMSECV), coefficient of determination (R2 ) and the deviation of Ratio of performance Deviation of the models (RPD). Exploratory data analysis through spectral treatments and principal component analysis (PCA), showed that the databases were similar to each other, and may be treated asa single model for each feed - mesquite pods and cactus. Evaluating the reference results, it was observed that the variation were similar to those reported in the literature. Comparing the preprocessing of samples, the performance ofthose developed with preprocessing (dried and ground) of the sample were more robust than those built with fresh samples. The NIRS Perten device performance similar to FOSS equipment, although the latter cover a larger spectral range and with lower readings intervals. NIR technology associate do multivariate techniques is reliable to predict the bromatological composition of mesquite pods and cactus.

New methodologies in CT perfusion and MRI analysis to develop cancer imaging biomarkers

Quantitative imaging in oncology aims at developing imaging biomarkers for diagnosis and prediction of cancer aggressiveness and therapy response before any morphological change become visible. This Thesis exploits Computed Tomography perfusion (CTp) and multiparametric Magnetic Resonance Imaging (mpMRI) for investigating diverse cancer features on different organs. I developed a voxel-based image analysis methodology in CTp and extended its use to mpMRI, for performing precise and accurate analyses at single-voxel level. This is expected to improve reproducibility of measurements and cancer mechanisms’ comprehension and clinical interpretability. CTp has not entered the clinical routine yet, although its usefulness in the monitoring of cancer angiogenesis, due to different perfusion computing methods yielding unreproducible results. Instead, machine learning applications in mpMRI, useful to detect imaging features representative of cancer heterogeneity, are mostly limited to clinical research, because of results’ variability and difficult interpretability, which make clinicians not confident in clinical applications. In hepatic CTp, I investigated whether, and under what conditions, two widely adopted perfusion methods, Maximum Slope (MS) and Deconvolution (DV), could yield reproducible parameters. To this end, I developed signal processing methods to model the first pass kinetics and remove any numerical cause hampering the reproducibility. In mpMRI, I proposed a new approach to extract local first-order features, aiming at preserving spatial reference and making their interpretation easier. In CTp, I found out the cause of MS and DV non-reproducibility: MS and DV represent two different states of the system. Transport delays invalidate MS assumptions and, by correcting MS formulation, I have obtained the voxel-based equivalence of the two methods. In mpMRI, the developed predictive models allowed (i) detecting rectal cancers responding to neoadjuvant chemoradiation showing, at pre-therapy, sparse coarse subregions with altered density, and (ii) predicting clinically significant prostate cancers stemming from the disproportion between high- and low- diffusivity gland components.

Development of a Computer-Based methodology for tolerance selection and optimization applied to the automotive sector

The design optimization of industrial products has always been an essential activity to improve product quality while reducing time-to-market and production costs. Although cost management is very complex and comprises all phases of the product life cycle, the control of geometrical and dimensional variations, known as Dimensional Management (DM), allows compliance with product and process requirements. Hence, the tolerance-cost optimization becomes the main practice to provide an effective application of Design for Tolerancing (DfT) and Design to Cost (DtC) approaches by enabling a connection between product tolerances and associated manufacturing costs. However, despite the growing interest in this topic, a profitable application in the industry of these techniques is hampered by their complexity: the definition of a systematic framework is the key element to improving design optimization, enhancing the concurrent use of Computer-Aided tools and Model-Based Definition (MBD) practices. The present doctorate research aims to define and develop an integrated methodology for product/process design optimization, to better exploit the new capabilities of advanced simulations and tools. By implementing predictive models and multi-disciplinary optimization, a Computer-Aided Integrated framework for tolerance-cost optimization has been proposed to allow the integration of DfT and DtC approaches and their direct application for the design of automotive components. Several case studies have been considered, with the final application of the integrated framework on a high-performance V12 engine assembly, to achieve both functional targets and cost reduction. From a scientific point of view, the proposed methodology provides an improvement for the tolerance-cost optimization of industrial components. The integration of theoretical approaches and Computer-Aided tools allows to analyse the influence of tolerances on both product performance and manufacturing costs. The case studies proved the suitability of the methodology for its application in the industrial field, providing the identification of further areas for improvement and refinement.