• Refine Query
  • Source
  • Publication year
  • to
  • Language
  • 10
  • 8
  • 2
  • 2
  • 1
  • 1
  • 1
  • Tagged with
  • 27
  • 7
  • 6
  • 5
  • 4
  • 4
  • 3
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • 2
  • About
  • The Global ETD Search service is a free service for researchers to find electronic theses and dissertations. This service is provided by the Networked Digital Library of Theses and Dissertations.
    Our metadata is collected from universities around the world. If you manage a university/consortium/country archive and want to be added, details can be found on the NDLTD website.
1

Spare parts provisioning decision support model for long lead time spares

Aulakh, Amit 06 1900 (has links)
Large corporations have a significant amount of working capital tied into the acquisition and storage of spare parts. In the industry, spare parts inventory policies and strategies are often developed in isolation from reliability centered maintenance practices – this results in significant wasteful direct and indirect cost attached to spare parts management for the equipment operator. This report will focus on developing a methodology for minimizing lifecycle indirect and direct cost that comes from storing long lead time spares. A combined Monte-Carlo and Genetic Algorithm based optimization approach to finding the optimal spare parts storage strategy is proposed. In this study, the indirect and direct cost of having a spare part in the storage facility will be balanced against the cost of lost opportunity that results from decreased availability - a consequence of not having the required spare part available when an equipment failure event occurs. The results of this study present the benefits of optimizing long lead time spares through a joint Monte-Carlo & Genetic Algorithm based approach. / Engineering Management
2

Spare parts provisioning decision support model for long lead time spares

Aulakh, Amit Unknown Date
No description available.
3

Análisis de influencia bajo inferencia bayesiana en evaluaciones escolares de altas consecuencias

Christiansen Trujillo, Andrés Guillermo 30 July 2018 (has links)
La presente investigación estudia una metodología para la detección de observaciones atípicas mediante un análisis de influencia bajo la perspectiva de la inferencia bayesiana. Se utiliza la medida de phi-divergencia y el estimador de Monte Carlo, derivado de ésta, trabajados previamente por Peng y Dey (1995), para el cálculo de las divergencias Kullback-Leibler, distancia rectilínea y ji-cuadrado. Además, en el presente trabajo se busca realizar este análisis de influencia en evaluaciones de altas consecuencias (evaluaciones cuyos resultados tienen un alto impacto en la vida de los estudiantes o docentes). El estudio de simulación revela que es posible recuperar observaciones previamente distorsionadas como atípicas. Finalmente, se aplica la metodología a una evaluación realizada por el Ministerio de Educación. Esta aplicación revela que la metodología estudiada es capaz de identificar escuelas con resultados no esperados dadas sus condiciones y resultados anteriores. / Tesis
4

Model for estimation of time and cost based on risk evaluation applied on tunnel projects

Isaksson, Therese January 2002 (has links)
No description available.
5

Model for estimation of time and cost based on risk evaluation applied on tunnel projects

Isaksson, Therese January 2002 (has links)
No description available.
6

Search, Inference and Opponent Modelling in an Expert-Caliber Skat Player

Long, Jeffrey Richard Unknown Date
No description available.
7

Clinical implementation of a Monte Carlo-based platform for the validation of stereotactic and intensity-modulated radiation therapy

Wagner, Antoine 27 August 2020 (has links) (PDF)
En radiothérapie, le niveau de précision de la dose délivrée au patient au cours de son traitement est d’une importance essentielle dans l’évolution vers une amélioration de la qualité et de la cohérence des données de suivi. L’une des premières étapes vers un système de support à la décision clinique (Clinical-Decision Support System CDSS) est la reconstruction précise de cette dose délivrée, en prenant en compte les nombreux facteurs pouvant générer des déviations significatives entre la dose planifiée visualisée à l’écran par l’utilisateur et la dose réellement accumulée lors des séances de traitement. Ces facteurs incluent les variations de débit de l’accélérateur, les incertitudes d’étalonnage, de calcul de dose, les mouvements du patient et des organes, etc.L’objectif de cette étude est d’implémenter et tester une plate-forme de calcul Monte Carlo pour la validation des systèmes Cyberknife et Tomothérapie installés au Centre Oscar Lambret. L’étude d’un détecteur dédié aux petits faisceaux (la chambre d’ionisation microLion) est également incluse, ce détecteur étant particulièrement adapté aux mesures sur le système Cyberknife.Le contexte et les concepts théoriques sont introduits dans les deux premiers chapitres. Dans le troisième chapitre, la modélisation Monte Carlo du Cyberknife et du détecteur microLion est détaillée. La quatrième partie inclut la description de la plate-forme Moderato et de son module d’évaluation. Dans le dernier chapitre, la modélisation du dernier modèle de Cyberknife (M6) équipé d’un collimateur multi-lames est décrite. Une nouvelle technique est également introduite dans le but d’accélérer la recherche des paramètres du faisceau d’électrons pour un modèle Monte Carlo, permettant une intégration plus simple et automatisée de nouveaux appareils dans Moderato. / In radiation therapy, the accuracy of the dose delivered to the patient during the course of treatment is of great importance to progress towards improved quality and coherence of the outcome data. One of the first steps to evolve towards a Clinical-Decision Support System (CDSS) is to be able to accurately reconstruct that delivered dose, taking into account the range of factors that can potentially generate significant differences between the planned dose visualized on the screen of the dosimetrist, and the actually delivered dose accumulated during the treatment sessions. These factors include accelerator output variations, commissioning uncertainties, dose computation errors, patient and organ movement, etc.The objective of this work is to implement and test a Monte Carlo platform for the validation of the Cyberknife and Tomotherapy systems installed at Centre Oscar Lambret. A study of a small field-dedicated detector (the microLion ionization chamber) is also included, this detector being particularly suited for measurements on the Cyberknife system.The context and theoretical concepts are introduced in the first two chapters. In the third chapter, the Monte Carlo modelling of the Cyberknife and microLion detector is detailed. The fourth part includes the description of the Monte Carlo platform Moderato and its evaluation module. In the final chapter, the modelling of the latest MLC-equipped Cyberknife model (the M6) is described. A new technique is also introduced to accelerate the optimization of the beam electron parameters of a Monte Carlo model, thus allowing for an easier and more automated use of the Moderato system. / Doctorat en Sciences biomédicales et pharmaceutiques (Médecine) / info:eu-repo/semantics/nonPublished
8

Modeling and Simulation of Polarization Mode Dispersion and Polarization Dependent Loss

Reimer, Michael January 2007 (has links)
Novel theoretical formulations and efficient simulation methods for polarization-mode dispersion (PMD) and polarization-dependent loss (PDL) that are directly applicable to optical network design are developed. In particular, a formalism based upon the Magnus expansion is advanced for the determination of the frequency evolution of the Mueller matrix in terms of increasing orders of PMD and PDL. Several previous models of polarization evolution are shown to be specializations of this more general formalism. A least-squares algorithm that extracts PMD and PDL coefficients from repeated measurements of the output Stokes vector of an optical system for a random set of input polarization states is introduced and subsequently applied to the rapid experimental determination of the probability density of the differential group delay of a fiber-squeezer based PMD emulator. The applicability of Clifford algebra and Pade-approximant techniques to the efficient simulation of the wavelength dependence of PMD and PDL is also discussed.
9

Simulação Monte Carlo e estudos analíticos em processos de solidificação / Monte Carlo simulation and analytic studies of phase transition

Silva, Marco Antonio Alves da 17 August 1987 (has links)
A função partição do modelo SOS é analisada, mostrando heuristicamente que acima da transição de rugosidade, TR, este modelo se torna essencialmente um problema unidimensional. Os resultados para a rugosidade superficial e calor específico são comparados com uma simulação Monte Carlo de uma rede mxm (m=20, 40, 80, 100). Um novo modelo SOS unidimensional é introduzido o qual mostra um bom acordo para regiões de altas e baixas temperaturas com o modelo SOS bidimensional. Foi dedicada atenção especial na estimativa da temperatura de transição pelo método Monte Carlo, a qual foi estabelecida como sendo TR= 0.80. Através de um aprimoramento no método Monte Carlo, aplicado a um sistema contendo 224 discos duros, obteve-se o loop de van der Waals, indicando assim, uma coexistência de fases para densidades entre 1.29 &#60 &#964 &#60 1.36. Este resultado até então, só tinha sido obtido através da técnica dinâmica molecular para 870 discos / The partition function of the SOS model is analysed and it is shown heuristically that above the roughening transition temperature, TR, this model becomes essentially na one-dimensional problem. The results for the surface roughness and surface specific heat are compared with a Monte Carlo simulation of a mxm lattice (m= 20, 40, 80, 100). A new one-dimensional SOS model is introduced wich shows a good agreement with the ordinary two dimensional SOS model for low and high temperatures. Especial attention was paid for the Monte Carlo estimation of the roughening transition temperature which was stablished as being TR= 0.80. Through na improvement in the choice of the most representative configurations in the Monte Carlo procedure, applied to a system containing 224 hard disks, it was possible to obtain the van der Waals-type loop curve, showing then, a fase coexistence in a density range 1.29 &#60 &#964 &#60 1.36. This result has so far only been obtained by molecular dynamic procedure for 870 disks
10

Evaluación del desempeño sísmico de puentes continuos

Vargas Bejarano, Cesar Isidoro 22 September 2017 (has links)
En esta tesis se utilizan dos métodos para la evaluación sísmica de puentes. Los métodos usados son el análisis dinámico incremental y simulación de Montecarlo usando Redes neuronales artificiales para la generación de curvas de fragilidad. El análisis dinámico incremental arroja una base de datos bastante amplia. El tratamiento estadístico utilizado abarca conceptos tanto de estadística descriptiva como inferencial. Así se presentan histogramas, frecuencias relativas acumuladas, valores de centralización, dispersión etc. Desde el punto de vista poblacional se presentan los valores de media y proporción poblacional para muestras pequeñas. Para la media se usó el teorema del límite central con la distribución t-student y para la proporción la distribución normal. El segundo método es un proceso de simulación con Montecarlo usando redes neuronales artificiales. Montecarlo toma la muestra de manera aleatoria, debido a esto para obtener resultados confiables se necesitan muchas simulaciones que conllevaría a un costo numérico muy alto. Por ello se usó las redes neuronales artificiales como “reemplazo” del modelo estructural no lineal. Para lograr esto la red se “entreno” con una base de datos del modelo estructural. Para la regresión se utilizó una red supervisada tipo feedforward (red hacia adelante), con el algoritmo de entrenamiento backpropagation (retropropación). La conclusión del trabajo confirma que con la red neuronal artificial se obtienen errores aceptables demostrando que es un MÉTODO DE REGRESIÓN poderoso para sistemas no lineales. La metodología propuesta demostró ser un método de simulación práctico debido a que usa redes neuronales entrenadas para generar curvas de fragilidad. Esto debido a que las redes neuronales tienen un costo numérico menor a un análisis dinámico no lineal. / Tesis

Page generated in 0.0505 seconds