Este trabajo de tesis ha tenido un largo desarrollo temporal, que ha permitido abordar campos de aplicación de contextos tradicionalmente muy distantes, permitiendo tener una visión más amplia y fortaleciendo el trabajo realizado. Un agente inteligente es una entidad que actúa de manera racional intentando lograr un objetivo, para ello percibe su entorno, procesa estas percepciones y responde en su entorno. Pero la resolución de un problema complejo presenta grandes dificultades y restricciones, tanto de tiempo como de recursos, así como necesita gran cantidad de conocimiento. Esto hace que la dificultad de resolución de los problemas aumente o las haga inviables para un agente trabajando individualmente. En los sistemas multiagente un grupo de agentes colaborarán, combinando sus habilidades y su conocimiento, permitiendo la resolución del problema de manera conjunta. Los sistemas multiagente son cada vez más demandados y utilizados. Su aplicación la podremos encontrar en un amplio número de campos como robótica, Big Data, aplicaciones de transporte y logística, sistemas de información geográfica, balanceo de redes y móviles, diagnóstico médico, aplicaciones gráficas, juegos, así como en muchos otros campos. A lo largo de los años se ha trabajado mucho para facilitar el diseño y la construcción de sistemas multiagente, han aparecido diversas arquitecturas, metodologías y herramientas, incluso se ha creado un estándar, el estándar IEEE-FIPA (Foundation for Intelligent Physical Agents). Pero se ha dejado la implementación a bajo nivel en manos de los equipos de desarrollo, los cuales han ido creando soluciones adhoc para cada uno de sus problemas. En esta tesis se presenta una arquitectura versátil que permita avanzar en el diseño y construcción de sistemas multiagente. Para ello proponemos una arquitectura híbrida multinivel y una metodología que nos facilita la creación de estos sistemas de una manera rápida, con bajo coste y de alta calidad. Para validar la arquitectura y demostrar su versatilidad la hemos aplicado a tres contextos de diferentes: • Edificios inteligentes: en sistema multiagente para mejorar el ahorro energético en un hotel, donde podremos observar una de las ventajas de la arquitectura: la integración de elementos de bajo nivel a través de la integración de elementos IoT (Internet of Things). • Robótica: en sistema de enjambre robótico para la detección de señales de radiofrecuencia, que demostrará la flexibilidad, escalabilidad y robustez de la arquitectura al aplicarse en un contexto de múltiples robots y de simulación. • Sistemas biomédicos: en una herramienta tecnológica para medir y registrar movimiento de hiperactividad (ADHD), donde podemos observar la versatilidad de la arquitectura a través de un contexto totalmente diferente, el de diagnósticos en salud.

No description available.

As more studies accumulate on the impact of diabetes mellitus on the central nervous system, they resound with the same conclusion - diabetes has a detrimental effect on cognition regardless of the presence of comorbidities. Less consistent however, are the specific mental processes wherein these declines are noticeable, and the structural changes that accompany these reductions in mental capacity. From global atrophy to changes in the volume of gray and white matter, to conflicting results regarding the effects of hypo- and hyperglycemic states on the development of the hippocampus, the studies display a variety of results. The goal of this research is to link the structural and compositional changes occurring in the diabetic brain with the clinical and behavioral findings highlighted in the literature, as well as to explore the potential mechanisms behind the pathologic brain state of diabetic encephalopathy. Using diabetic (OVE26) and non-diabetic wild type (FVB) mice as models, differences in the number of hippocampal neurons in the dentate gyrus, and cornu ammonis areas 1,2, and 3 were investigated through Nissl staining. Neurodegeneration was confirmed in those cells determined to be hyperchromatic in the diabetic model through staining with Fluoro-Jade C. Finally, the presence of progenitor cells in the hippocampus was compared in the diabetic and non-diabetic models using Musashi-1 antibodies, to determine whether neurogenesis in these areas is affected by diabetes. These experiments were performed to better understand the effect of DM on learning and memory, and could potentially explain the linkage between diabetes mellitus and the increased prevalence of Alzheimer’s disease, vascular dementia, and depression in this subset of the population.

No description available.

Cette thèse présente un modèle unique du MASG (Modèle d’Architecture du Smart Grid) en considérant l 'état de l’art des différentes directives de recherche du smart grid. Le système hybride de génération d'énergie active marine-hydrogène a été modélisé pour représenter la couche de composants du MASG. Le système intègre l'électrolyseur à membrane d’échange de proton (à l’échelle de méga watt) et les systèmes de piles à combustible en tant que composants principaux du bilan énergétique. La batterie LiFePO4 est utilisée pour couvrir la dynamique rapide de l'énergie électrique. En outre, la thèse analyse le système de gestion de l'énergie centralisé et décentralisé. Le système multi-agents représente le paradigme du système décentralisé. La plate-forme JADE est utilisée pour développer le système multi-agents, en raison de son domaine d'application général, de ses logiciels à licence libre, de son interface avec MATLAB et de sa calculabilité avec les standards de la Fondation des Agents Physiques Intelligentes. Le système de gestion d'énergie basé sur JADE équilibre l'énergie entre la génération (système de conversion d'énergie marine-courant) et la demande (profil de charge résidentielle) pendant les modes de fonctionnement autonome et connecté au réseau. Le modèle proposé du MASG peut être considéré comme une étude de cas pilote qui permet l'analyse détaillée et les applications des différentes directions de recherche du smart grid. / This thesis presents a unique model of the SGAM (Smart Grid Architecture Model) with considering the state of the art of the different research directions of the smart grid and. The hybrid marine-hydrogen active power generation system has been modeled to represent the component layer of the SGAM. The system integrates the MW scale PEM electrolyzer and fuel cell systems as the main energy balance components. The LiFePO4 battery is used to cover the fast dynamics of the electrical energy. Moreover, the thesis analyzes the centralized and the decentralized energy management system. The MAS (Multi-Agent Systems) represents the paradigm of the decentralized system. The JADE platform is used to develop the MAS due to its general domain of application, open source and free license software, interface with MATLAB and the computability with the FIPA (Foundation of Intelligent Physical Agent) standards. The JADE based energy management system balances the energy between the generation (marine-current energy conversion system) and the demand side (residential load profile) during the stand-alone and the grid-connected modes of operation. The proposed model of the SGAM can be considered as a pilot case study that enables the detailed analysis and the applications of the different smart grid research directions.

It is impossible to remove methylmercury (MeHg) from biological systems because MeHg is found throughout our environment in many fresh and salt water fish. The consumption of fish is important to human nutrition and health. The mechanism of MeHg neurotoxicity must be understood to minimize adverse exposure consequences. The dissertation objective was to: 1) compare mechanisms of MeHg neurotoxicity between animals exposed as adults and those exposed during gestation, and 2) develop an in vitro test model of in vivo MeHg exposure. Total mercury (Hg) levels in tissue / cells were determined by combustion / trapping / atomic absorption. Cell death was determined by Fluoro-Jade histochemical staining and activated caspase 3 immunohistochemistry for in vivo studies, and Trypan blue exclusion, lactate dehydrogenase activity, and cytotoxicity assays for in vitro studies. Mitochondrial membrane potential (MMP), intracellular calcium ion concentration ([Ca2+]i), and production of reactive oxygen species (ROS) were determined using fluorescence microscopy or microplate reader assays. Young adult C57Bl/6 mice were exposed to a total dose of 0, 1.0, or 5.0 mg/kg body weight MeHg divided over postnatal days (P)35 to 39. Pregnant female mice were exposed to a total does of 0, 0.1, or 1.0 mg/kg body weight MeHg divided over gestational days (G)8 to 18. SY5Y cells were exposed to 0, 0.01, 0.1, or 1.0 ?M MeHg or HgCl2 for 24, 48, or 72 hours. Total Hg in brains of young adult mice, mouse pups, and SY5Y cells accumulated in a dose-dependent manner. Cell death increased in SY5Y cells exposed to the highest concentrations of MeHg and HgCl2 used in this study. Cell death increased in the molecular and granule cerebellar cell layers of young adult mice exposed to the highest doses of MeHg used in this study. P0 mouse pups showed no increase in cell death within the cerebellum following MeHg exposure. Cerebella of mice at P10 exhibited decreased dying cells only in the external germinal layer. Low concentrations of MeHg affected MMP in both in vivo and in vitro studies, but did not result in decreased MMP typically associated with higher MeHg concentrations. [Ca2+]i was increased throughout the in vivo experiments in an age- , sexand brain region-dependent manner. Generation of ROS was decreased in both in vivo and in vitro studies with both the MeHg and HgCl2 (in vitro) treatments. In summary, low and moderate MeHg exposure, both in vivo and in vitro, altered mitochondrial function, Ca2+ homeostasis, and ROS differently than what is reported in the literature for higher MeHg exposure concentrations. SY5Y cells were sensitive to low-levels of MeHg and HgCl2 and responded similarly to cells in the whole animal studies, thus making SY5Y cells realistic candidates for mechanistic MeHg studies. Cell culture and whole animal neuronal functional studies at chronic low-level MeHg exposure are limited. These data suggest that low-levels of MeHg may affect neuronal function. Therefore, further chronic low-level MeHg neuronal functional studies are warranted.