From catalytic studies in surface science, it has been shown that the catalytic activity is dependent on the type of metal facet used. Nanocrystals of different shapes have different facets. This raises the possibility that the use of metal nanoparticles of different shapes could catalyze different reactions with different efficiencies. The catalytic activity is found to correlate with the fraction of surface atoms located on the corners and edges of the tetrahedral, cubic, and spherical platinum nanoparticles. It is observed that for nanoparticles of comparable size, the tetrahedral nanoparticles have the highest fraction of surface atoms located on the corners and edges and also have the lowest activation energy, making them the most catalytically active. Nanoparticles have a high surface-to-volume ratio, which makes them attractive to use compared to bulk catalytic materials. However, their surface atoms are also very active due to their high surface energy. As a result, it is possible that the surface atoms are so active that their size and shape could change during the course of their catalytic function. It is found that dissolution of corner and edge atoms occurs for both the tetrahedral and cubic platinum nanoparticles during the full course of the mild electron transfer reaction and that there is a corresponding change in the activation energy in which both kinds of nanoparticles strive to behave like spherical nanoparticles. When spherical palladium nanoparticles are used as catalysts for the Suzuki reaction, it is found that the nanoparticles grow larger after the first cycle of the reaction due to the Ostwald ripening process since it is a relatively harsh reaction due to the need to reflux the reaction mixture for 12 hours at 100 oC. When the tetrahedral Pt nanoparticles are used to catalyze this reaction, the tetrahedral nanoparticles transform to spherical ones, which grow larger during the second cycle. In addition, studies on the effect of the individual reactant have also provided clues to the surface catalytic process that is taking place. In the case of the electron transfer reaction, the surface catalytic process involves the thiosulfate ions binding to the nanoparticle surface and reacting with the hexacyanoferrate (III) ions in solution. In the case of the Suzuki reaction, the surface catalytic mechanism of the Suzuki reaction involves the phenylboronic acid binding to the nanoparticle surface and reacting with iodobenzene via collisional processes.

Pattern recognition in urban areas is one of the most challenging issues in classifying satellite remote sensing data. Parametric pixel-by-pixel classification algorithms tend to perform poorly in this context. This is because urban areas comprise a complex spatial assemblage of disparate land cover types - including built structures, numerous vegetation types, bare soil and water bodies. Thus, there is a need for more powerful spectral pattern recognition techniques, utilizing pixel-by-pixel spectral information as the basis for automated urban land cover detection. This paper adopts the multi-layer perceptron classifier suggested and implemented in [5]. The objective of this study is to analyse the performance and stability of this classifier - trained and tested for supervised classification (8 a priori given land use classes) of a Landsat-5 TM image (270 x 360 pixels) from the city of Vienna and its northern surroundings - along with varying the training data set in the single-training-site case. The performance is measured in terms of total classification, map user's and map producer's accuracies. In addition, the stability with initial parameter conditions, classification error matrices, and error curves are analysed in some detail. (authors' abstract) / Series: Discussion Papers of the Institute for Economic Geography and GIScience

L'hybridation des sources de puissance dans le domaine des applications embarquées s'est imposée comme une solution adéquate pour répondre aux législations environnementales et atteindre une meilleure efficacité énergétique. Toutefois, le choix dans le dimensionnement des composants et la stratégie de commande doivent répondre à un cahier des charges, souvent complexe et hétérogène, tout en limitant les coûts du système. La résolution de ce problème d'optimisation incluant de nombreuses variables peut s'avérer complexe à cause des non-linéarités présentes dans le problème formulé. Il faut donc disposer d'outils de résolution efficaces et capables de fournir une solution fiable. Dans cette thèse, nous proposons une méthode d'optimisation globale pour le dimensionnement et la commande optimale de véhicules hybrides basée sur l'optimisation combinatoire, et en particulier sur la programmation linéaire en nombres entiers (PLNE). A partir d'un problème d'optimisation non linéaire, le problème initial est reformulé en une multitude de sous-problèmes linéaires en nombres entiers sur lesquels un algorithme de Branch & Bound parallèle est exécuté. Afin de résoudre des problèmes de grande taille, un second algorithme basé sur le Branch & Cut est développé. Cette méthode est déployée pour l'étude d'un système d'alimentation hybride d'une mini-excavatrice électrique. Le problème d'optimisation, dans lequel des contraintes énergétiques et des contraintes de vieillissement sont implantées, est évalué suivant différents paramètres du cahier des charges. Enfin, cette approche est également appliquée pour l'optimisation de trajectoires d'un système multi-actionneur synchronisés. / Hybridization of power sources for embedded applications becomes an interesting solution to respect environmental legislation and achieve a higher energy efficiency. However, the choice for components sizing and the energy management strategy need to meet specifications while reducing costs. To solve this optimization problems including several types of variables can be complex because of non linearities included in the formulated problem. Therefore the use of effective solving tools, able to provide a reliable solution, is required. In this thesis, a global optimization method is proposed for the design and the optimal control of hybrid vehicles based on combinatorial optimization, particularly on integer linear programming. From a non-linear optimization problem, the initial problem is reformulated into a multitude of integer linear sub-problems for which a parallel Branch & Bound algorithm is executed. In order to solve large-scale problems, a second algorithm based on the Branch & Cut is developed. This method is used for the study of a hybrid power supply system of a mini-excavator electric. The optimization problem, where energy constraints and aging constraints are implemented, is evaluated according to several parameters and specifications. Finally, this approach is also applied for the optimization of trajectories for a synchronized multi-actuators system.