<p>Uncertainty exists everywhere in scientific and engineering applications. To avoid potential risk, it is critical to understand the impact of uncertainty on a system by performing uncertainty quantification (UQ) and reliability analysis (RA). However, the computational cost may be unaffordable using current UQ methods with high-dimensional input. Moreover, current UQ methods are not applicable when numerical data and image data coexist. </p> <p>To decrease the computational cost to an affordable level and enable UQ with special high dimensional data (e.g. image), this dissertation develops three UQ methodologies with high dimensionality of input space. The first two methods focus on high-dimensional numerical input. The core strategy of Methodology 1 is fixing the unimportant variables at their first step most probable point (MPP) so that the dimensionality is reduced. An accurate RA method is used in the reduced space. The final reliability is obtained by accounting for the contributions of important and unimportant variables. Methodology 2 addresses the issue that the dimensionality cannot be reduced when most of the variables are important or when variables equally contribute to the system. Methodology 2 develops an efficient surrogate modeling method for high dimensional UQ using Generalized Sliced Inverse Regression (GSIR), Gaussian Process (GP)-based active learning, and importance sampling. A cost-efficient GP model is built in the latent space after dimension reduction by GSIR. And the failure boundary is identified through active learning that adds optimal training points iteratively. In Methodology 3, a Convolutional Neural Networks (CNN) based surrogate model (CNN-GP) is constructed for dealing with mixed numerical and image data. The numerical data are first converted into images and the converted images are then merged with existing image data. The merged images are fed to CNN for training. Then, we use the latent variables of the CNN model to integrate CNN with GP to quantify the model error using epistemic uncertainty. Both epistemic uncertainty and aleatory uncertainty are considered in uncertainty propagation. </p> <p>The simulation results indicate that the first two methodologies can not only improve the efficiency but also maintain adequate accuracy for the problems with high-dimensional numerical input. GSIR with active learning can handle the situations that the dimensionality cannot be reduced when most of the variables are important or the importance of variables are close. The two methodologies can be combined as a two-stage dimension reduction for high-dimensional numerical input. The third method, CNN-GP, is capable of dealing with special high-dimensional input, mixed numerical and image data, with the satisfying regression accuracy and providing an estimate of the model error. Uncertainty propagation considering both epistemic uncertainty and aleatory uncertainty provides better accuracy. The proposed methods could be potentially applied to engineering design and decision making. </p>

La propuesta tiene como objetivo mejorar el proceso productivo y la gestión de compras para lograr una mayor eficiencia y aumentar la competitividad de la empresa en el mercado. En ese sentido, se desarrolló e implementó un plan de optimización para el proceso de producción que comenzó con el análisis de la causa raíz, con ello, se determinó que existe una gestión de procesos inadecuada y una planificación y control de producción ineficientes. Por lo tanto, la propuesta consiste en el uso de herramientas lean para todo el proceso operativo en la gestión de compras, y la estandarización con 5s para mediciones y controles en las operaciones. El uso de estas herramientas como un conjunto, permite reducir los tiempos de ciclo, controlar las operaciones y sistematizarlas para que funcionen de forma agrupada y automática. Por último, queremos ser parte de esta nueva era digital y desarrollarla a un más alto nivel, ya que con ayuda del diseño se tienen las estrategias definidas y el plan completo para efectuar la mejora, encontramos una de las mejores plataformas de aplicaciones empresariales open source que ayuda a potenciar el negocio para obtener un sistema que coordine sus procesos de manera automática, el cual certifique el buen funcionamiento de la entidad y permita prestar un servicio de calidad. Como resultado de la aplicación, hubo una reducción en el tiempo de ciclo del 8%, la tasa de reprocesamiento en un 10% y los costos de producción se redujeron en 11.2% durante el primer semestre. / The proposal aims to improve the performance of the production process and the management of purchases to achieve greater efficiency and increase the competitiveness of the company in the market. In that sense, an optimization plan was developed and implemented for the production process of the company that began with the analysis of the root cause, it was determined that there is inadequate process management and inadequate production planning and control. Therefore, the proposal consists in the use of lean tools such as lean logistics for the entire operational process in purchasing management and standardization with 5s for measurements and controls in the operations of the production area. The use of these tools as a set, allows reducing cycle times, controlling operations and systematizing them so that they work in a grouped and automatic way. As a result of the application, there was a reduction in the cycle time of 8%, the reprocess rate in 10% and delays in the delivery time of the finished products by 25% less. Finally, production costs were reduced by 11.2% for the first semester. / Trabajo de Suficiencia Profesional

No description available.

La simulation numérique directe des couches limites turbulentes nécessite une résolution spatiale élevée, en particulier proche des parois afin de capturer les différentes échelles de longueur inhérentes à la turbulence. Pour réduire le coût de calcul lié à la haute résolution spatiale, nous étudions les techniques qui éliminent la région de paroi en utilisant une condition aux limites artificielle aux bords d’un domaine de calcul réduit. Une stratégie de reconstruction basée sur la POD est mise en œuvre afin de générer des conditions aux limites synthétiques permettant de récupérer les caractéristiques spatiales et dynamiques correctes de la turbulence dans la région centrale du canal. La condition aux limites a été testée à deux hauteurs différentes. Les statistiques du canal réduit ont été comparées à celles du canal complet prise comme valeur de référence. Une bonne concordance est observée pour le profil de vitesse moyenne et la contrainte de cisaillement turbulente. La procédure s’est avérée robuste à la hauteur à laquelle la condition aux limites est imposée et permet une réduction drastique des coûts de calcul. / Direct Numerical Simulation of turbulent boundary layers requires a high spatial resolution specially near the wall to capture the different length scales inherent of turbulence. To reduce the computing cost related to the high spatial resolution, we investigate techniques that bypass the wall region by using an artificial boundary condition at the bound of a reduced computational domain. A reconstruction strategy based on POD is implemented in order to generate synthetic boundary conditions such that the correct spatial characteristics and dynamics of the turbulence are recovered in the core region of the channel. The boundary condition was tested at two different heights. Statistics in the reduced channel were compared with those of a reference case. A good agreement is observed for the mean velocity profile and the turbulent shear stress. The procedure was found to be robust with the boundary height and results in a drastic computational cost reduction.

This thesis has attempted to reduce costs associated with wear parts in test equipment at Ericsson’s master and industrialization center (MIC) in Kista. Two different cases have been pursued, and each has been designed to follow a, for the purpose, adapted version of the design research methodology (DRM). The adapted model was presented and its use in both cases was additionally evaluated. Both cases were based on products that already have been launched and are used in production. In order to avoid expensive changes to surrounding equipment, incremental product development was chosen in both cases. The first case focused on test equipment used in the production of digital units for radio base stations. After the wear parts of this test equipment were examined, it was found that a plastic detail was manufactured through milling and as a result was quite expensive. The plastic detail was redesigned so that it could be manufactured using the injection molding process instead. This resulted in an estimated cost reduction of 48 %. The second case focused on wear parts of test fixtures used in the production of radio filters, also used in Ericsson’s radio base stations. A more process-oriented view was adopted in this case and parameters believed to be causing added wear such as positional and angular misalignment were examined further. Tolerance chains for each wear part in the assembled test fixture were calculated to determine the degrees of possible misalignment. The resulting data did however not correlate with each wear part’s individual wear interval, and no concrete cost reductive improvements could be delivered. Instead suggestions which can lead to further cost reductions in the long term were presented. The model used to develop improvements for these two cases was found to have helped the process in both cases, however further research of this model is recommended in order to determine if it is suitable for these types of projects in general

No description available.

The need for annotated corpora in a variety of different types of research grows constantly. Unfortunately creating annotated corpora is frequently cost-prohibitive due the number of person-hours required to create the corpus. This project investigates one solution that helps to reduce the cost of creating annotated corpora through the use of a new user interface which includes a specially built framework and component for annotating part-of-speech information and the implementation of a dictionary. This project reports on a user study performed to determine the effect of dictionaries with different levels of coverage on a part-of-speech annotation task. Based on a pilot study with thirty-three participants the analysis shows that a part-of-speech tag dictionary with greater than or equal to 60% coverage helps to improve the time required to complete the part-of-speech annotation task while maintaining high levels of accuracy.

Yes

Cette recherche porte sur l’étude de l’impact du partage de plusieurs informations simultanément dans différentes configurations de chaînes logistiques. Celui-ci s’exprime en termes de réduction du coût logistique par rapport au scénario de non partage d’informations et est évalué au niveau de la chaîne logistique dans sa globalité, mais également au niveau de chaque coût élémentaire de chaque maillon. Le choix des informations partagées est basé sur leur importance dans la gestion de la chaîne logistique ainsi que sur leur provenance (amont et aval) pour permettre d’instaurer des mécanismes d’échange et de négociation entre les différents maillons. Notre approche est développée dans le cadre d’une chaîne logistique série mono-produit composée d’un fournisseur, d’un entrepôt, d’un détaillant et de clients. Nous avons développé un modèle mathématique permettant l’évaluation des gains de ce partage d’informations dans le cadre d’une décision décentralisée. Ce premier modèle nous a permis d’étudier les impacts du partage de ces informations dans des circonstances proches de la réalité suite à l’injection de perturbations même dans le cas de partage. L’étude expérimentale réalisée a permis de mettre en évidence la nécessité de mettre en place des mécanismes incitatifs de coopération entre les différents maillons de la chaîne logistique lorsque les gains ne bénéficient de manière quasi exclusive qu’à un seul des maillons ou s’avèrent particulièrement déséquilibrés, mécanismes visant à aboutir à des équilibres gagnant-gagnant entre les participants. Nous avons ensuite considéré le cas d’une décision décentralisée avec partage de deux informations simultanément. Cette partie de notre travail est réalisée sur une chaîne logistique divergente mono-produit intégrant plusieurs détaillants. L’affectation par l’entrepôt des produits aux détaillants est réalisée au prorata des commandes détaillants. Nous avons proposé de mettre en place des mécanismes de coopération nécessaires pour garantir un gain équilibré entre les différents maillons dans les différents scénarios de partage. Nous avons considéré par la suite une décision centralisée au niveau de l’entrepôt avec une affectation de produits minimisant le coût logistique de toute la chaîne. Pour terminer, notre approche a été étendue à l’étude de la répartition des gains entre détaillants dans différentes configurations de demande et ce dans les différents scenarii étudiés. Ceci nous a permis d’identifier les bénéficiaires de ce partage d’informations dans différents contextes. Nous avons conclu par le fait qu’une décision décentralisée avec partage d’informations est proche en termes de performance de celle obtenue dans un contexte centralisé et préconisé de poursuivre cette étude comparative en prenant en compte l’ensemble des coûts et bénéfices de ces deux modes de pilotage / We study the impact of sharing several informations simultaneously in different configurations of supply chains. The valuation is expressed in terms of logistics costs reduction compared to the scenario of non-information sharing. We have evaluated the gain in the overall supply chain, but also at each elementary cost of a particular link. The choice of the shared information is based on their importance in the supply chain management as well as their source (upstream and downstream) in order to establish different exchange mechanisms and negotiation between various links. We have studied the sharing of one, two and three informations simultaneously. Our approach is developed in the context of a mono-product serial supply chain consisting of one supplier, one warehouse, one retailer and customers. We have developed a mathematical model in order to evaluate the gains provided by the sharing of different kinds of information in the context of decentralized decision. In this model, we have injected a perturbation in the shared information in order to be more realistic. This experimental study has highlighted the necessity to establish incentive cooperation mechanisms between the various links in the supply chain, if the gains are benefic exclusively to only one of the links or are particularly unbalanced. These mechanisms are intended in order to lead to a win-win balance between participants. In the second part of our work, we have considered the case of a decentralized decision with sharing simultaneously two pieces of information. This part is carried out in the context of a mono-product and a divergent supply chain integrating several retailers. The quantities allocation from the warehouse to retailers is made in a prorata assignment policy based on retailers’ orders. We have noted that the sharing in its various forms has led to significant reduction cost for the overall supply chain. We have considered later the case of a centralized decision at the warehouse with an assignment of products in order to minimize the supply chain cost. Finally, our approach has been extended to the study of the distribution of gains between retailers for different customer demand configurations. This has allowed us to identify the beneficiaries of this information sharing in different studied contexts. The obtained gains are rarely balanced between links; this resulted in imposing the establishment of cooperative mechanisms incentive for sharing. We have concluded by the fact that a decentralized decision with information sharing is close in performance to a centralized context and we plan to continue this comparative study by taking into account all costs and benefits of these two pilot modes.

Ao longo dos anos, diferentes ferramentas vem sendo utilizadas pelos administradores de rede para realizar as tarefas de gerenciamento (por exemplo, protocolos de gerência e soluções de monitoramento de rede). Dentre tais ferramentas, a presente tese foca em Visualização de Informações (ou simplesmente InfoVis). Essencialmente, entende-se que o objetivo final dessas ferramentas de gestão é diminuir a complexidade e, consequentemente, otimizar o trabalho diário dos administradores. Assim, eles podem melhorar sua produtividade, o que incide diretamente na redução de custos. Com base nesse pressuposto, esta tese tem como objetivo investigar como promover a adoção de técnicas InfoVis pelos administradores de rede, com foco em melhorar produtividade e diminuir custos. A percepção chave é que, na maioria dos casos, os administradores de rede não são habilitados no domínio InfoVis. Desse modo, a escolha por adotar técnica InfoVis requer a imersão em campo desconhecido, podendo gerar, assim, um risco elevado nos indicadores de produtividade e custos. Em essência, essa tese argumenta que o emprego de técnicas InfoVis pelos administradores pode ser muito trabalhosa, despendendo um montante muito significativo de tempo, o que leva a diminuir produtividade e, consequentemente, eleva os custos de gerenciamento. Focando essa questão, é apresentada uma proposta para promover adoção de técnicas InfoVis, pelo encorajamento do reuso. Argumenta-se que os conceitos e princípios de reuso propostos e padronizados pelo campo da engenharia de software podem ser adaptados e empregados, uma vez que a construção de visualizações (ou seja, o projeto e desenvolvimento) é, primariamente, uma tarefa de desenvolvimento de software. Assim, a avaliação da proposta apresentada nesta tese utiliza o método Common Software Measurement International Consortium (COSMIC) Functional Size Measurement (FSM), o qual permite estimar o dimensionamento de software através de pontos por função. A partir deste método, torna-se então possível a estimativa de esforço e, consequentemente, produtividade e custos. Os resultados mostram a viabilidade e eficácia da abordagem proposta (em termos de produtividade e custos), bem como os benefícios indiretos que o reuso sistemático pode fornecer quando da adoção de visualizações para auxílio nas tarefas de gerenciamento de redes. / Throughout the years, several tools have being used by network administrators to accomplish the management tasks (e.g., management protocols and network monitoring solutions). Among such tools, this thesis focuses on Information Visualization one (a.k.a InfoVis). Mainly, it is understood that the ultimate goal of these management tools is to decrease the complexity and, consequently, optimize the everyday work of administrators. Thus, they can increase their productivity, which leads to the cost reduction. Based on this assumption, this thesis aims at investigating how to promote the adoption of InfoVis techniques by network administrators, focusing on enhancing productivity and lowering costs. The key insight is that, in most cases, network administrators are unskilled on InfoVis. Therefore, the choice to adopt visualizations can require an immersion into the unknown that can be too risky regarding productivity and cost. In essence, this thesis argues that the employment of InfoVis techniques by administrators can be very laborious by spending a significant amount of effort that decreases their productivity and, consequently, increases the management costs. To overcome this issue, an approach to promote the adoption of InfoVis techniques by encouraging their reuse is introduced. It is argued that the concepts and principles of software reuse proposed and standardized in the software engineering field can be adapted and employed once the building up of visualizations (i.e., the design and development) can be defined primarily as a software development task. So, the evaluation of the proposal introduced in this thesis employs the Common Software Measurement International Consortium (COSMIC) Functional Size Measurement (FSM) method that measures software sizing through Function Points (FP). From this method, it was possible estimating effort and, consequently, productivity and costs. Results show the feasibility and effectiveness of the proposed approach (in terms of productivity and cost) as well as some indirect benefits that the systematic reuse can provide in the adoption of InfoVis techniques to assist in the management tasks.