Determining the number of clusters in categorical data

Cluster analysis for categorical data has been an active area of research. A well-known problem in this area is the determination of the number of clusters, which is unknown and must be inferred from the data. In order to estimate the number of clusters, one often resorts to information criteria, such as BIC (Bayesian information criterion), MML (minimum message length, proposed by Wallace and Boulton, 1968), and ICL (integrated classification likelihood). In this work, we adopt the approach developed by Figueiredo and Jain (2002) for clustering continuous data. They use an MML criterion to select the number of clusters and a variant of the EM algorithm to estimate the model parameters. This EM variant seamlessly integrates model estimation and selection in a single algorithm. For clustering categorical data, we assume a finite mixture of multinomial distributions and implement a new EM algorithm, following a previous version (Silvestre et al., 2008). Results obtained with synthetic datasets are encouraging. The main advantage of the proposed approach, when compared to the above referred criteria, is the speed of execution, which is especially relevant when dealing with large data sets.

Simulação de processos de adsorção recorrendo ao Aspen Chromatography

As operações de separação por adsorção têm vindo a ganhar importância nos últimos anos, especialmente com o desenvolvimento de técnicas de simulação de leitos móveis em colunas, tal como a cromatografia de Leito Móvel Simulado (Simulated Moving Bed, SMB). Esta tecnologia foi desenvolvida no início dos anos 60 como método alternativo ao processo de Leito Móvel Verdadeiro (True Moving Bed, TMB), de modo a resolver vários dos problemas associados ao movimento da fase sólida, usuais nestes métodos de separação cromatográficos de contracorrente. A tecnologia de SMB tem sido amplamente utilizada em escala industrial principalmente nas indústrias petroquímica e de transformação de açúcares e, mais recentemente, na indústria farmacêutica e de química fina. Nas últimas décadas, o crescente interesse na tecnologia de SMB, fruto do alto rendimento e eficiente consumo de solvente, levou à formulação de diferentes modos de operação, ditos não convencionais, que conseguem unidades mais flexíveis, capazes de aumentar o desempenho de separação e alargar ainda mais a gama de aplicação da tecnologia. Um dos exemplos mais estudados e implementados é o caso do processo Varicol, no qual se procede a um movimento assíncrono de portas. Neste âmbito, o presente trabalho foca-se na simulação, análise e avaliação da tecnologia de SMB para dois casos de separação distintos: a separação de uma mistura de frutose-glucose e a separação de uma mistura racémica de pindolol. Para ambos os casos foram considerados e comparados dois modos de operação da unidade de SMB: o modo convencional e o modo Varicol. Desta forma, foi realizada a implementação e simulação de ambos os casos de separação no simulador de processos Aspen Chromatography, mediante a utilização de duas unidades de SMB distintas (SMB convencional e SMB Varicol). Para a separação da mistura frutose-glucose, no quediz respeito à modelização da unidade de SMB convencional, foram utilizadas duas abordagens: a de um leito móvel verdadeiro (modelo TMB) e a de um leito móvel simulado real (modelo SMB). Para a separação da mistura racémica de pindolol foi considerada apenas a modelização pelo modelo SMB. No caso da separação da mistura frutose-glucose, procedeu-se ainda à otimização de ambas as unidades de SMB convencional e Varicol, com o intuito do aumento das suas produtividades. A otimização foi realizada mediante a aplicação de um procedimento de planeamento experimental, onde as experiências foram planeadas, conduzidas e posteriormente analisadas através da análise de variância (ANOVA). A análise estatística permitiu selecionar os níveis dos fatores de controlo de modo a obter melhores resultados para ambas as unidades de SMB.

Simulation model for driving dynamics, energy use and power supply

Dissertação para obtenção do grau de Mestre em Engenharia Electrotécnica Ramo de Energia

The generalized multiprocessor periodic resource interface model for hierarchical multiprocessor scheduling

Composition is a practice of key importance in software engineering. When real-time applications are composed it is necessary that their timing properties (such as meeting the deadlines) are guaranteed. The composition is performed by establishing an interface between the application and the physical platform. Such an interface does typically contain information about the amount of computing capacity needed by the application. In multiprocessor platforms, the interface should also present information about the degree of parallelism. Recently there have been quite a few interface proposals. However, they are either too complex to be handled or too pessimistic.In this paper we propose the Generalized Multiprocessor Periodic Resource model (GMPR) that is strictly superior to the MPR model without requiring a too detailed description. We describe a method to generate the interface from the application specification. All these methods have been implemented in Matlab routines that are publicly available.

Analysis of temperature time-series: embedding dynamics into the MDS method

Global warming and the associated climate changes are being the subject of intensive research due to their major impact on social, economic and health aspects of the human life. Surface temperature time-series characterise Earth as a slow dynamics spatiotemporal system, evidencing long memory behaviour, typical of fractional order systems. Such phenomena are difficult to model and analyse, demanding for alternative approaches. This paper studies the complex correlations between global temperature time-series using the Multidimensional scaling (MDS) approach. MDS provides a graphical representation of the pattern of climatic similarities between regions around the globe. The similarities are quantified through two mathematical indices that correlate the monthly average temperatures observed in meteorological stations, over a given period of time. Furthermore, time dynamics is analysed by performing the MDS analysis over slices sampling the time series. MDS generates maps describing the stations’ locus in the perspective that, if they are perceived to be similar to each other, then they are placed on the map forming clusters. We show that MDS provides an intuitive and useful visual representation of the complex relationships that are present among temperature time-series, which are not perceived on traditional geographic maps. Moreover, MDS avoids sensitivity to the irregular distribution density of the meteorological stations.

Hybrid adaptive control of a dragonfly model

Dragonflies show unique and superior flight performances than most of other insect species and birds. They are equipped with two pairs of independently controlled wings granting an unmatchable flying performance and robustness. In this paper, it is presented an adaptive scheme controlling a nonlinear model inspired in a dragonfly-like robot. It is proposed a hybrid adaptive (HA) law for adjusting the parameters analyzing the tracking error. At the current stage of the project it is considered essential the development of computational simulation models based in the dynamics to test whether strategies or algorithms of control, parts of the system (such as different wing configurations, tail) as well as the complete system. The performance analysis proves the superiority of the HA law over the direct adaptive (DA) method in terms of faster and improved tracking and parameter convergence.

Image quality and dose analysis in PA chest x-ray: comparison between AEC and manual mode acquisition using the 10 kVp rule

Purpose: To compare image quality and effective dose when the 10 kVp rule is applied with manual and AEC mode in PA chest X-ray. Methods and Materials: A total of 68 images (with and without lesions) were acquired of an anthropomorphic chest phantom in a Wolverson Arcoma X-ray unit. The images were evaluated against a reference image using image quality criteria and the 2 alternative forced choice (2 AFC) method by five radiographers. The effective dose was calculated using PCXMC software using the exposure parameters and DAP. The exposure index (lgM) was recorded. Results: Exposure time decreases considerably when applying the 10 kVp rule in manual mode (50%-28%) compared to AEC mode (36%-23%). Statistical differences for effective dose between several AEC modes were found (p=0.002). The effective dose is lower when using only the right AEC ionization chamber. Considering image quality, there are no statistical differences (p=0.348) between the different AEC modes for images with no lesions. Using a higher kVp value the lgM values will also increase. The lgM values showed significant statistical differences (p=0.000). The image quality scores did not present statistically significant differences (p=0.043) for the images with lesions when comparing manual with AEC modes. Conclusion: In general, the dose is lower in the manual mode. By using the right AEC ionising chamber the effective dose will be the lowest in comparison to other ionising chambers. The use of the 10 kVp rule did not affect the detectability of the lesions.

Automatic tunning of Okumura-Hata model on railway communications

This paper presents the Genetic Algorithms (GA) as an efficient solution for the Okumura-Hata prediction model tuning on railways communications. A method for modelling the propagation model tuning parameters was presented. The algorithm tuning and validation were based on real networks measurements carried out on four different propagation scenarios and several performance indicators were used. It was shown that the proposed GA is able to produce significant improvements over the original model. The algorithm developed is currently been used on real GSM-R network planning process for an enhanced resources usage.

A straightforward method to obtain the cohesive laws of bonded joints under mode I loading

A simple procedure to measure the cohesive laws of bonded joints under mode I loading using the double cantilever beam test is proposed. The method only requires recording the applied load–displacement data and measuring the crack opening displacement at its tip in the course of the experimental test. The strain energy release rate is obtained by a procedure involving the Timoshenko beam theory, the specimen’s compliance and the crack equivalent concept. Following the proposed approach the influence of the fracture process zone is taken into account which is fundamental for an accurate estimation of the failure process details. The cohesive law is obtained by differentiation of the strain energy release rate as a function of the crack opening displacement. The model was validated numerically considering three representative cohesive laws. Numerical simulations using finite element analysis including cohesive zone modeling were performed. The good agreement between the inputted and resulting laws for all the cases considered validates the model. An experimental confirmation was also performed by comparing the numerical and experimental load–displacement curves. The numerical load–displacement curves were obtained by adjusting typical cohesive laws to the ones measured experimentally following the proposed approach and using finite element analysis including cohesive zone modeling. Once again, good agreement was obtained in the comparisons thus demonstrating the good performance of the proposed methodology.

Supplier's selection model based on an empirical study

The problem of selecting suppliers/partners is a crucial and important part in the process of decision making for companies that intend to perform competitively in their area of activity. The selection of supplier/partner is a time and resource-consuming task that involves data collection and a careful analysis of the factors that can positively or negatively influence the choice. Nevertheless it is a critical process that affects significantly the operational performance of each company. In this work, there were identified five broad selection criteria: Quality, Financial, Synergies, Cost, and Production System. Within these criteria, it was also included five sub-criteria. After the identification criteria, a survey was elaborated and companies were contacted in order to understand which factors have more weight in their decisions to choose the partners. Interpreted the results and processed the data, it was adopted a model of linear weighting to reflect the importance of each factor. The model has a hierarchical structure and can be applied with the Analytic Hierarchy Process (AHP) method or Value Analysis. The goal of the paper it's to supply a selection reference model that can represent an orientation/pattern for a decision making on the suppliers/partners selection process

Feasibility of the extended finite element method for the simulation of composite bonded joints

Adhesive-bonding for the unions in multi-component structures is gaining momentum over welding, riveting and fastening. It is vital for the design of bonded structures the availability of accurate damage models, to minimize design costs and time to market. Cohesive Zone Models (CZM’s) have been used for fracture prediction in structures. The eXtended Finite Element Method (XFEM) is a recent improvement of the Finite Element Method (FEM) that relies on traction-separation laws similar to those of CZM’s but it allows the growth of discontinuities within bulk solids along an arbitrary path, by enriching degrees of freedom. This work proposes and validates a damage law to model crack propagation in a thin layer of a structural epoxy adhesive using the XFEM. The fracture toughness in pure mode I (GIc) and tensile cohesive strength (sn0) were defined by Double-Cantilever Beam (DCB) and bulk tensile tests, respectively, which permitted to build the damage law. The XFEM simulations of the DCB tests accurately matched the experimental load-displacement (P-d) curves, which validated the analysis procedure.

Strength prediction of adhesively-bonded scarf repairs in composite structures under bending

This work reports on the experimental and numerical study of the bending behaviour of two-dimensional adhesively-bonded scarf repairs of carbon-epoxy laminates, bonded with the ductile adhesive Araldite 2015®. Scarf angles varying from 2 to 45º were tested. The experimental work performed was used to validate a numerical Finite Element analysis using ABAQUS® and a methodology developed by the authors to predict the strength of bonded assemblies. This methodology consists on replacing the adhesive layer by cohesive elements, including mixed-mode criteria to deal with the mixed-mode behaviour usually observed in structures. Trapezoidal laws in pure modes I and II were used to account for the ductility of the adhesive used. The cohesive laws in pure modes I and II were determined with Double Cantilever Beam and End-Notched Flexure tests, respectively, using an inverse method. Since in the experiments interlaminar and transverse intralaminar failures of the carbon-epoxy components also occurred in some regions, cohesive laws to simulate these failure modes were also obtained experimentally with a similar procedure. A good correlation with the experiments was found on the elastic stiffness, maximum load and failure mode of the repairs, showing that this methodology simulates accurately the mechanical behaviour of bonded assemblies.

Buckling strength of adhesively-bonded single and double-strap repairs on carbon-epoxy structures

This work reports on an experimental and finite element method (FEM) parametric study of adhesively-bonded single and double-strap repairs on carbon-epoxy structures under buckling unrestrained compression. The influence of the overlap length and patch thickness was evaluated. This loading gains a particular significance from the additional characteristic mechanisms of structures under compression, such as fibres microbuckling, for buckling restrained structures, or global buckling of the assembly, if no transverse restriction exists. The FEM analysis is based on the use of cohesive elements including mixed-mode criteria to simulate a cohesive fracture of the adhesive layer. Trapezoidal laws in pure modes I and II were used to account for the ductility of most structural adhesives. These laws were estimated for the adhesive used from double cantilever beam (DCB) and end-notched flexure (ENF) tests, respectively, using an inverse technique. The pure mode III cohesive law was equalled to the pure mode II one. Compression failure in the laminates was predicted using a stress-based criterion. The accurate FEM predictions open a good prospect for the reduction of the extensive experimentation in the design of carbon-epoxy repairs. Design principles were also established for these repairs under buckling.

Development of biocompatible and “smart” porous structures using CO2-assisted processes

Dissertação apresentada para a obtenção do grau de Doutor em Engenharia Química, especialidade Engenharia da Reacção Química, pela Universidade Nova de Lisboa, Faculdade de Ciências e Tecnologia

The daily and hourly energy consumption and load forecasting using artifitial neural network method: a case study using a set of 93 households in Portugal

It is important to understand and forecast a typical or a particularly household daily consumption in order to design and size suitable renewable energy systems and energy storage. In this research for Short Term Load Forecasting (STLF) it has been used Artificial Neural Networks (ANN) and, despite the consumption unpredictability, it has been shown the possibility to forecast the electricity consumption of a household with certainty. The ANNs are recognized to be a potential methodology for modeling hourly and daily energy consumption and load forecasting. Input variables such as apartment area, numbers of occupants, electrical appliance consumption and Boolean inputs as hourly meter system were considered. Furthermore, the investigation carried out aims to define an ANN architecture and a training algorithm in order to achieve a robust model to be used in forecasting energy consumption in a typical household. It was observed that a feed-forward ANN and the Levenberg-Marquardt algorithm provided a good performance. For this research it was used a database with consumption records, logged in 93 real households, in Lisbon, Portugal, between February 2000 and July 2001, including both weekdays and weekend. The results show that the ANN approach provides a reliable model for forecasting household electric energy consumption and load profile. © 2014 The Author.