Process models are always associated with uncertainty, due to either inaccurate model structure or inaccurate identification. If left unaccounted for, these uncertainties can significantly affect the model-based decision-making. This thesis addresses the problem of model-based optimization in the presence of uncertainties, especially due to model structure error. The optimal solution from standard optimization techniques is often associated with a certain degree of uncertainty and if the model-plant mismatch is very significant, this solution may have a significant bias with respect to the actual process optimum. Accordingly, in this thesis, we developed new strategies to reduce (1) the variability in the optimal solution and (2) the bias between the predicted and the true process optima. Robust optimization is a well-established methodology where the variability in optimization objective is considered explicitly in the cost function, leading to a solution that is robust to model uncertainties. However, the reported robust formulations have few limitations especially in the context of nonlinear models. The standard technique to quantify the effect of model uncertainties is based on the linearization of underlying model that may not be valid if the noise in measurements is quite high. To address this limitation, uncertainty descriptions based on the Bayes’ Theorem are implemented in this work. Since for nonlinear models the resulting Bayesian uncertainty may have a non-standard form with no analytical solution, the propagation of this uncertainty onto the optimum may become computationally challenging using conventional Monte Carlo techniques. To this end, an approach based on Polynomial Chaos expansions is developed. It is shown in a simulated case study that this approach resulted in drastic reductions in the computational time when compared to a standard Monte Carlo sampling technique. The key advantage of PC expansions is that they provide analytical expressions for statistical moments even if the uncertainty in variables is non-standard. These expansions were also used to speed up the calculation of likelihood function within the Bayesian framework. Here, a methodology based on Multi-Resolution analysis is proposed to formulate the PC based approximated model with higher accuracy over the parameter space that is most likely based on the given measurements. For the second objective, i.e. reducing the bias between the predicted and true process optima, an iterative optimization algorithm is developed which progressively corrects the model for structural error as the algorithm proceeds towards the true process optimum. The standard technique is to calibrate the model at some initial operating conditions and, then, use this model to search for an optimal solution. Since the identification and optimization objectives are solved independently, when there is a mismatch between the process and the model, the parameter estimates cannot satisfy these two objectives simultaneously. To this end, in the proposed methodology, corrections are added to the model in such a way that the updated parameter estimates reduce the conflict between the identification and optimization objectives. Unlike the standard estimation technique that minimizes only the prediction error at a given set of operating conditions, the proposed algorithm also includes the differences between the predicted and measured gradients of the optimization objective and/or constraints in the estimation. In the initial version of the algorithm, the proposed correction is based on the linearization of model outputs. Then, in the second part, the correction is extended by using a quadratic approximation of the model, which, for the given case study, resulted in much faster convergence as compared to the earlier version. Finally, the methodologies mentioned above were combined to formulate a robust iterative optimization strategy that converges to the true process optimum with minimum variability in the search path. One of the major findings of this thesis is that the robust optimal solutions based on the Bayesian parametric uncertainty are much less conservative than their counterparts based on normally distributed parameters.

Orientador: Secundino Soares Filho / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Eletrica e de Computação / Made available in DSpace on 2018-08-08T14:25:59Z (GMT). No. of bitstreams: 1 Marques_ThyagoCarvalho_D.pdf: 9101920 bytes, checksum: 5cf6202ab737e2bcab996c2704ede5ec (MD5) Previous issue date: 2006 / Resumo: Este trabalho apresenta uma política para a operação energética do Sistema Nacional (SIN), baseada na representação individualizada das usinas hidrelétricas e termelétricas, na representação detalhada das características de operação dessas usinas, e na representação indireta da estocasticidade das vazões através demodelo de previsão. Essas características diferem a política operativa proposta daquela em vigor no setor elétrico brasileiro, que se baseia na modelagem equivalente de sistemas hidrotérmicos interligados. Diversos estudos de caso são apresentados com o intuito de avaliar o desempenho dessa polÂ'ıtica, desde o caso de sistemas hidrotérmicos simples com uma única usina até o caso completo do SIN sob condições reais. Para esse último caso, um procedimento de ajuste da operação foi implementado para assegurar o atendimento do mercado com as restrições de intercâmbio entre os subsistemas. Análises de sensibilidade foram realizadas em relação a restrições de intercâmbio, inflexibilidade das usinas termelétricas, a importância de algumas usinas hidrelétricas e mercado / Abstract: This work presents an operational policy for the mid term hydrothermal scheduling of the Brazilian Interconnected Power System (BIPS) based on the individual representation of the hydro and thermo power plants, the detailed representation of the plantsâ?¿ features and the indirect representation of the inflow stochasticity through a forecasting model. These characteristics differs the operational policy which is in use nowadays in BIPS, which is based on four interconnected subsystems represented by composite reservoirs. Several case studies are presented to evaluate the performance of the operational policy, from the simple case with a single hydro plant system to the complete case of the whole BIPS under real conditions. For this last case an operational adjustment procedure was implemented to assure the market attendance considering interchange constraints between the subsystems. Sensitivity analyses have been done with respected to interchange, minimum thermal generation, market, importance of hydro plants and system dimension / Doutorado / Energia Eletrica / Doutor em Engenharia Elétrica

No description available.

L’un des challenges les plus importants des pays développés ou émergents est de soutenir l’innovation dans les PME. Mais la question centrale ici, est de savoir si l’innovation dépend exclusivement de la disponibilité des ressources. La capacité à innover ne dispose pas d'un cadre de référence pour son évaluation, en raison principalement de sa complexité et sa nature abstraite. Sans les indicateurs appropriés, il est impossible d'établir les objectifs et de déterminer les efforts et les ressources nécessaires pour améliorer la capacité d'innovation. Le laboratoire ERPI a développé une approche d’évaluation de cette capacité privilégiant les pratiques internes qui favorisent l’innovation dans les entreprises (IIP, Indice d’Innovation Potentielle) basée sur des méthodes d’analyse multicritère. Mais une fois obtenu un résultat, quels sont les points sur lesquels une entreprise peut jouer afin d’améliorer son système d'innovation ? Dans cette thèse, nous proposons un modèle d'optimisation qui réplique la procédure d'évaluation de la capacité à innover utilisée par l'IIP et qui nous permettra de générer un plan d'action pour améliorer la capacité à innover des PME. Le modèle considère différents facteurs (la situation particulière de l'entreprise, les efforts économiques associés à la stratégie proposée, etc.) afin de générer le plan d'action le plus approprié aux besoins de chaque PME. Ainsi, trois stratégies potentielles sont proposées : améliorer le classement actuel avec un minimum d'effort, maximiser les économies de ressources, maximiser l’IIP avec un budget défini. Des scénarii sont simulés afin de tester les résultats de la mise en œuvre des différentes stratégies / One of the most important challenges in the current industrial policy of developed and emerging countries is to support innovation in SMEs, issue which has been approach as a problem only dependent on the availability of resources. Innovation capability does not have an appropriate framework of evaluation, mainly because of its complexity and abstract nature. Without proper indicators, it is not possible to establish objectives and identify the efforts and resources necessary for its improvement. In response to this, the ERPI laboratory has developed an evaluation approach of innovation capability, taking into account internal practices to promote innovation. A field of knowledge has been developed in this direction allowing for the proposal of a novel metrology, the Potential Innovation index (PII), based on multi-criteria analysis methods for studying the dynamics within companies. In this thesis, we propose an optimization model that replicates the evaluation procedure of innovation capability to generate an action plan, i.e., the model returns concrete points in which companies can work to improve its innovation system. The model considers factors such as the particular situation of the company, the economic efforts associated with the proposed strategy, the goals of the entrepreneur, etc. Three optimization strategies are proposed: access to a higher category with a minimum of effort, maximize resource efficiency and maximize innovation performance with a defined budget. Scenarios are simulated to test the results of the implementation of the different strategies

Multi-objective evolutionary algorithms have gained a lot of atten- tion in the recent years. They have proven to be among the best multi-objective optimizers and have been used in many industrial ap- plications. However, their usability is hindered by the large number of evaluations of the objective functions they require. These can be expensive when solving practical tasks. In order to reduce the num- ber of objective function evaluations, surrogate models can be used. These are a simple and fast approximations of the real objectives. In this work we present the results of research made between the years 2009 and 2013. We present a multi-objective evolutionary algo- rithm with aggregate surrogate model, its newer version, which also uses a surrogate model for the pre-selection of individuals. In the next part we discuss the problem of selection of a particular type of model. We show which characteristics of the various models are im- portant and desirable and provide a framework which combines sur- rogate modeling with meta-learning. Finally, in the last part, we ap- ply multi-objective optimization to the problem of hyper-parameters tuning. We show that additional objectives can make finding of good parameters for classifiers faster. 1

Le nombre de véhicules circulant sur les routes dans le monde est passé de 500 millions en 1986 à 1 milliard de véhicules en 2010. Un tel réseau routier immense a apporté un certain confort à de nombreux conducteurs mais d'un autre coté il a représenté environs 1,24 millions d'accidents en 2010. Avec ces chiffres vient l'augmentation du niveau des émissions de CO2 et des milliards d'heures perdues dans les embouteillages. Avec le progrès et le développement des technologies sans fil ces dernières années, les réseaux véhiculaires ont rapidement évolués. Beaucoup de constructeurs automobiles et de pays ont intégré et imaginé différentes applications dans les réseaux véhiculaires : des applications de sécurité routières, d'autres pour l'information sur le trafic, ainsi que des applications de divertissement. Cette dernière catégorie d'applications se distingue des autres par la quantité et le volume de données échangées, et constitue un réel problème dû à la nature des communications sans fil véhiculaires. Dans cette thèse, nous traitons le problème de la dissémination d'information dans les systèmes véhiculaires distribués. Notre objectif principal est de proposer des solutions pratiques et réalisables pour diffuser les données dans des environnements véhiculaires réalistes. Le concept de la diffusion des données est large et significatif. Dans ce contexte, nous nous référons à la diffusion de données chaque fois qu'il y a une certaine quantité de données qui doit être répartie sur un réseau distribué sans fil. Les données sont envoyées vers plusieurs destinations (même à tous les nœuds du réseau dans certain cas) qui sont intéressés pour récupérer la totalité ou une partie des informations générées. A titre d'exemples, nous pouvons considérer les applications de partage de fichiers, la diffusion de fichiers multimédia et ainsi de suite. Nous abordons dans cette thèse le problème de dissémination de données dans les réseaux véhiculaires, plus précisément la dissémination de grands volumes de données tel que le contenu multimédia. Avec l'évolution au cours de ces dernières années des réseaux véhiculaires notamment grâce au support de nombreux constructeurs automobiles (e.x. Toyota, Nissan, BMW, Ford, etc.), la quantité de données échangée a augmenté de manière exponentielle au fil des années. Malheureusement malgré l'avancement des technologies de communications sans fil, l'échange de grands volumes de données dans des réseaux véhiculaires reste un réel défi, à cause des caractéristiques de communications sans fil et celles spécifiques aux réseaux véhiculaires. / The number of vehicles on the roads in the world increased from 500 million in 1986 to 1 billion vehicles in 2010. Such a huge road network has brought some comfort to many drivers but on the other hand it represented around 1.24 million accidents in 2010. With the progress and development of wireless technologies these recent years, vehicular networks quickly evolved. Many car manufacturers and countries have built and imagined different applications in vehicular networks: road safety applications, others for traffic information, and entertainment applications. This last category of applications is distinguished from others by the quantity and volume of exchanged data, and represents a real problem due to the lossy nature of vehicular communications. In this thesis, we address the problem of data dissemination in distributed vehicular systems. Our main objective is to provide practical and workable solutions to disseminate data in realistic vehicular environments. The concept of data dissemination is large and significant. In this context, we call data dissemination each time there is a certain amount of data which must be distributed on in wireless network. The data is sent to multiple destinations who are interested to recover all or part of the information. As examples, we can consider the file sharing applications, streaming multimedia files and so on. We discuss in this thesis the data dissemination problem in vehicular networks, specifically the dissemination of large volumes of data such as content distribution. With the evolution in recent years of vehicular networks thanks to many car manufacturers support (eg Toyota, Nissan, BMW, Ford, etc.), the amount of exchanged data has increased exponentially over the years. Unfortunately, despite the advancement of wireless communications technologies, exchange large volumes of data in vehicular networks remains a challenge because of wireless communications features and those specific to vehicular networks.

The impact of climate changes on various sectors of human activity and preparation of mitigation and adaptation measures are becoming an important subject of global policy, thus also the policy of the Czech Republic. According to the research results, the impact of climate changes on minimum discharges is very significant and indicates a threat to storage function of water sources, without adequate accumulation for period of drought. One of the ways of preventing or completely avoiding the appearance of these problems is the construction of new reservoirs or the re-evaluation of the size of the storage and protective capacity of existing reservoirs. In relation to climate changes and their impact on the water yield there is a growing need for the participation of operative control progressive methods of reservoir storage function today. The dissertation thesis is aimed at storage function of water management system. Aim of the dissertation work is to create program with graphical user interface, which allows solving the storage function of water management system. Program allows automatically set up optimization and simulation model of any water supply system and solve of optimal control and optimal development of the water supply system. The program was verified on the selected water supply subsystem.

Las industrias alimentarias afrontan complicados problemas de gestión de inventario, ya que los productos son perecederos, es decir, que en un tiempo determinado el producto ya no será comestible. El problema de la PYME del sector agroindustrial es la ineficiencia en el control de inventario en el almacén de productos terminados. Por ello, la importancia de este proyecto es optimizar el control de inventario de productos olivos. Además, el objetivo de la investigación es minimizar los costos de inventario, ya que actualmente el sobrecosto es de S/ 319 204 soles. Para ello, se establecieron los indicadores de entregas perfectas, exactitud de los inventarios, rotura de stock y rotación de inventarios de manera que se logré un aumento en cada uno de ellos y la estandarización de manera eficiente. Por tal motivo, se va a implementar el método FIFO y la metodología PHVA (planear-hacer-verificar-actuar), también conocida como el ciclo Deming en la empresa agroindustrial con la finalidad de estandarizar los procesos de recepción, ubicación preparación y despacho en el almacén de productos terminados. / Food industries face complicated inventory management problems, since the products are perishable, that is, after a certain time the product will no longer be edible. The problem of the SME of the agro-industrial sector is the inefficiency in inventory control in the warehouse of finished products. Therefore, the importance of this project is to optimize the inventory control of olive products. In addition, the objective of the investigation is to minimize inventory costs, since currently the overcharge is 319 204 soles. For this, the indicators of perfect deliveries, inventory accuracy, stock break and inventory turnover were established so that an increase in each of them and standardization was achieved efficiently. For this reason, the FIFO method and the PHVA (plan-do-verify-act) methodology, also known as the Deming cycle, will be implemented in the agro-industrial company in order to standardize the reception, location, preparation and dispatch processes in the finished goods warehouse. / Trabajo de investigación

Магистерская диссертация состоит из введения, двух глав, заключения, библиографического списка, приложений. В теоретической части рассмотрены теоретические аспекты внутрифирменное обучение сотрудников, теория и практика осуществления в сфере оказания услуг, разработана модель оптимизации внутрифирменного обучения современной организации .В практической части даются общая характеристика исследуемого предприятия, анализ его кадрового состава, системы управления персоналом и основных составляющих модели оптимизации внутрифирменного обучения линейных сотрудников. На основе полученных данных разработаны рекомендации по совершенствованию текущей модели внутрифирменного обучения сотрудников. В заключении подведены итоги в соответствии с поставленными задачами. / The master's thesis consists of an introduction, two chapters, a conclusion, a bibliographic list, and appendices. In the theoretical part, the theoretical aspects of in-house training of employees, the theory and practice of implementation in the field of service provision are considered, a model for optimizing in-house training of a modern organization is developed. In the practical part, a general description of the enterprise under study, an analysis of its personnel, personnel management system and the main components of the model for optimizing the internal training of line employees are given. Based on the data obtained, recommendations were developed to improve the current model of in-house employee training. In conclusion, the results are summarized in accordance with the tasks set.

The Florida Showcase Green Envirohome (FSGE) incorporates many green technologies. FSGE is built to meet or exceed 12 green building guidelines and obtain 8 green building certificates. The two-story 3292 ft2 home is a "Near Zero-Loss Home", "Near Zero-Energy Home", "Near Zero-Runoff Home", and "Near Zero-Maintenance Home". It is spawned from the consumer-driven necessity to build a home resistant to hurricanes, tornadoes, floods, fire, mold, termites, impacts, and even earthquakes given up to 500% increase in insurance premiums in natural disaster zones, the dwindling flexibility and coverage of insurance policies, and rising energy, water and maintenance costs (FSGE 2008). The FSGE captures its stormwater runoff from the green roof, metal roof and wood decking area and routes it to the sustainable water cistern. Graywater from the home (after being disinfected using ozone) is also routed to the sustainable water cistern. This water stored in the sustainable water cistern is used for irrigation of the green roof, ground level landscape, and for toilet flushing water. This study was done in two phases. During phase one, only stormwater runoff from the green roof, metal roof and wood decking area is routed to the sustainable water cistern. Then, during phase two, the water from the graywater system is added to the sustainable water cistern. The sustainable water cistern quality is analyzed during both phases to determine if the water is acceptable for irrigation and also if it is suitable for use as toilet flushing water. The water quality of the sustainable cistern is acceptable for irrigation. The intent of the home is to not pollute the environment, so as much nutrients as possible should be removed from the wastewater before it is discharged into the groundwater. Thus, the FSGE design is to evaluate a new on-site sewage treatment and disposal (OSTD) system which consists of a sorption media labeled as Bold and GoldTM filtration media. The Bold and GoldTM filtration media is a mixture of tire crumb and other materials. This new OSTD system has sampling ports through the system to monitor the wastewater quality as it passes through. Also, the effluent wastewater quality is compared to that of a conventional system on the campus of the University of Central Florida. The cost-benefit optimization model focused on designing a residential home which incorporated a green roof, cistern and graywater systems. This model had two forms, the base model and the grey linear model. The base model used current average cost of construction of materials and installation. The grey model used an interval for the cost of construction materials and green roof energy savings. Both models included a probabilistic term to describe the rainfall amount. The cost and energy operation of a typical Florida home was used as a case study for these models. Also, some of the parameters of the model were varied to determine their effect on the results. The modeling showed that the FSGE 4500 gallon cistern design was cost effective in providing irrigation water. Also, the green roof area could have been smaller to be cost effective, because the green roof cost is relatively much higher than the cost of a regular roof.