abstract: The energy consumption by public drinking water and wastewater utilities represent up to 30%-40% of a municipality energy bill. The largest energy consumption is used to operate motors for pumping. As a result, the engineering and control community develop the Variable Speed Pumps (VSPs) which allow for regulating valves in the network instead of the traditional binary ON/OFF pumps. Potentially, VSPs save up to 90% of annual energy cost compared to the binary pump. The control problem has been tackled in the literature as “Pump Scheduling Optimization” (PSO) with a main focus on the cost minimization. Nonetheless, engineering literature is mostly concerned with the problem of understanding “healthy working conditions” (e.g., leakages, breakages) for a water infrastructure rather than the costs. This is very critical because if we operate a network under stress, it may satisfy the demand at present but will likely hinder network functionality in the future. This research addresses the problem of analyzing working conditions of large water systems by means of a detailed hydraulic simulation model (e.g., EPANet) to gain insights into feasibility with respect to pressure, tank level, etc. This work presents a new framework called Feasible Set Approximation – Probabilistic Branch and Bound (FSA-PBnB) for the definition and determination of feasible solutions in terms of pumps regulation. We propose the concept of feasibility distance, which is measured as the distance of the current solution from the feasibility frontier to estimate the distribution of the feasibility values across the solution space. Based on this estimate, pruning the infeasible regions and maintaining the feasible regions are proposed to identify the desired feasible solutions. We test the proposed algorithm with both theoretical and real water networks. The results demonstrate that FSA-PBnB has the capability to identify the feasibility profile in an efficient way. Additionally, with the feasibility distance, we can understand the quality of sub-region in terms of feasibility. The present work provides a basic feasibility determination framework on the low dimension problems. When FSA-PBnB extends to large scale constraint optimization problems, a more intelligent sampling method may be developed to further reduce the computational effort. / Dissertation/Thesis / Masters Thesis Industrial Engineering 2018

Multistage stochastic optimization is used to solve many real-life problems where decisions are taken at multiple times, e.g., portfolio selection problems. Such problems need the definition of stochastic processes, which are usually approxim- ated by scenario trees. The choice of the size of the scenario trees is the result of a compromise between the best approximation and the possibilities of the com- puter technology. Therefore, once a master scenario tree has been generated, it can be needed to reduce its dimension in order to make the problem computation- ally tractable. In this thesis, we introduce several scenario reduction algorithms and we compare them numerically for different types of master trees. A simple portfolio selection problem is also solved within the study. The distance from the initial scenario tree, the computational time, and the distance between the optimal objective values and solutions are compared for all the scenario reduction algorithms. In particular, we adopt the nested distance to measure the distance between two scenario trees. 1

L'utilisation d'équipes de robots a pris de l'ampleur ces dernières années. Cela est dû aux avantages que peut offrir une équipe de robot par rapport à un robot seul pour la réalisation d'une même tâche. Cela s'explique aussi par le fait que ce type de plates-formes deviennent de plus en plus abordables et fiables. Ainsi, l'utilisation d'une équipe de véhicules aériens devient une alternative viable. Cette thèse se concentre sur le problème du déploiement d'une équipe de Micro-Véhicules Aériens (MAV) pour effectuer des missions de surveillance sur un terrain inconnu de morphologie arbitraire. Puisque la morphologie du terrain est inconnue et peut être complexe et non-convexe, les algorithmes standards ne sont pas applicables au problème particulier traité dans cette thèse. Pour y remédier, une nouvelle approche basée sur un algorithme d'optimisation cognitive et adaptatif (CAO) est proposée et évaluée. Une propriété fondamentale de cette approche est qu'elle partage les mêmes caractéristiques de convergence que les algorithmes de descente de gradient avec contraintes qui exigent une connaissance parfaite de la morphologie du terrain pour optimiser la couverture. Il est également proposé une formulation différente du problème afin d'obtenir une solution distribuée, ce qui nous permet de surmonter les inconvénients d'une approche centralisée et d'envisager également des capacités de communication limitées. De rigoureux arguments mathématiques et des simulations étendues établissent que l'approche proposée fournit une méthodologie évolutive et efficace qui intègre toutes les contraintes physiques particulières et est capable de guider les robots vers un arrangement qui optimise localement la surveillance. Finalement, la méthode proposée est mise en œuvre sur une équipe de MAV réels pour réaliser la surveillance d'un environnement extérieur complexe. / The use of multi-robot teams has gained a lot of attention in recent years. This is due to the extended capabilities that the teams offer compared to the use of a single robot for the same task. Moreover, as these platforms become more and more affordable and robust, the use of teams of aerial vehicles is becoming a viable alternative. This thesis focuses on the problem of deploying a swarm of Micro Aerial Vehicles (MAV) to perform surveillance coverage missions over an unknown terrain of arbitrary morphology. Since the terrain's morphology is unknown and it can be quite complex and non-convex, standard algorithms are not applicable to the particular problem treated in this thesis. To overcome this, a new approach based on the Cognitive-based Adaptive Optimization (CAO) algorithm is proposed and evaluated. A fundamental property of this approach is that it shares the same convergence characteristics as those of constrained gradient-descent algorithms, which require perfect knowledge of the terrain's morphology to optimize coverage. In addition, it is also proposed a different formulation of the problem in order to obtain a distributed solution, which allows us to overcome the drawbacks of a centralized approach and to consider also limited communication capabilities. Rigorous mathematical arguments and extensive simulations establish that the proposed approach provides a scalable and efficient methodology that incorporates any particular physical constraints and limitations able to navigate the robots to an arrangement that (locally) optimizes the surveillance coverage. The proposed method is finally implemented in a real swarm of MAVs to carry out surveillance coverage in an outdoor complex area.

A presente pesquisa realizou o estudo da confiabilidade aplicado à otimização da operação em tempo real de sistemas de abastecimento de água (SAA). Almeja-se que a otimização da operação em tempo real empregue técnicas que a tornem robusta, ou seja, que considerem as incertezas inerentes a um SAA real. Para tanto, é necessário associar ao modelo de otimização um previsor de demanda e um simulador hidráulico. O previsor produzirá estimativas de demandas futuras para o horizonte desejado, o qual alimentará o simulador, a fim de que sejam determinadas as estratégias operacionais otimizadas para atendimento das demandas previstas. Implementou-se o método de otimização AMALGAM (\"A Multialgorithm Genetically Adaptive Method\"), juntamente com as demais rotinas computacionais necessárias para integrar o simulador hidráulico (EPANET 2) e o previsor de demanda baseado na Rede Neural Dinâmica (DAN2). O modelo desenvolvido foi aplicado nos setores de abastecimento Eliana, Iguatemi e Martinez, os quais são operados pelo Departamento Autônomo de Água e Esgotos (DAAE) da cidade de Araraquara, SP. Os modelos das redes de água foram calibrados por meio de dados de vazão e carga de pressão coletados em campanhas de campo. As estratégias operacionais resultantes foram comparadas as operações praticadas pelo DAAE, resultando em reduções no custo do consumo de energia de 14%, 13% e 30% para os setores Eliana, Iguatemi e Martinez, respectivamente. / This research project proposes the study of reliability applied to real time optimization of operation of water supply network (WSN). It is desired to obtain robust real time optimization of operation through the use of adequate techniques which accounts the inherent uncertainty of a real WSN. To accomplish the task it is necessary to associate to the optimization model a demand forecaster and a hydraulic simulator. The forecaster will produce the future demand for the planning horizon to serve as input for the simulator, so it is possible to obtain the optimized operation to meet the predicted demand. It was implemented the AMALGAM (\"A Multialgorithm Genetically Adaptive Method\") to serve as optimization model as well as the necessary computational routine to link the EPANET hydraulic simulator as well as the demand forecaster based on DAN2. The developed model was applied to the sectors Eliana, Iguatemi and Martinez, which are part of the water system operated by the Autonomous Department of Water and Sewer (DAAE) of Araraquara, SP. The water network model was calibrated using data collected on field campaign to gather pressure and flow data. The optimized operation was compared to the operation from DAAE, resulting in reduction of energy consumption cost of 14%, 13% and 30% respectively for the sectors Eliana, Iguatemi and Martinez.

Electric power is one of the most critical parts of everyday life; from lighting, heating, and cooling homes to powering televisions and computers. The modern power grids face several challenges such as efficiency, sustainability, and reliability. Increase in electrical energy demand, distributed generations, integration of uncertain renewable energy resources, and demand side management are among the main underlying reasons of such growing complexity. Additionally, the elements of power systems are often vulnerable to failures because of many reasons, such as system limits, poor maintenance, human errors, terrorist/cyber attacks, and natural phenomena. One common factor complicating the operation of electrical power systems is the underlying uncertainties from the demands, supplies and failures of system components. Stochastic optimization approaches provide mathematical frameworks for decision making under uncertainty. It enables a decision maker to incorporate some knowledge of the uncertainty into the decision making process to find an optimal trade off between cost and risk. In this dissertation, we focus on application of three risk-averse approaches to power systems modeling and optimization. Particularly, we develop models and algorithms addressing the cost-effectiveness and reliability issues in power grids with integrations of renewable energy resources. First, we consider a unit commitment problem for centralized hydrothermal systems where we study improving reliability of such systems under water inflow uncertainty. We present a two-stage robust mixed-integer model to find optimal unit commitment and economic dispatch decisions against extreme weather conditions such as drought years. Further, we employ time series analysis (specifically vector autoregressive models) to construct physical based uncertainty sets for water inflow into the reservoirs. Since extensive formulation is impractical to solve for moderate size networks we develop an efficient Benders' decomposition algorithm to solve this problem. We present the numerical results on real-life case study showing the effectiveness of the model and the proposed solution method. Next, we address the cost effectiveness and reliability issues considering the integration of solar energy in distributed (decentralized) generation (DG) such as microgrids. In particular, we consider optimal placement and sizing of DG units as well as long term generation planning to efficiently balance electric power demand and supply. However, the intermittent nature of renewable energy resources such as solar irradiance imposes several difficulties in decision making process. We propose two-stage stochastic programming model with chance constraints to control the risk of load shedding (i.e., power shortage) in distributed generation. We take advantage of another time series modeling approach known as autoregressive integrated moving average (ARIMA) model to characterize the uncertain solar irradiance more accurately. Additionally, we develop a combined sample average approximation (SAA) and linearization techniques to solve the problem more efficiently. We examine the proposed framework with numerical tests on a radial network in Arizona. Lastly, we address the robustness of strategic networks including power grids and airports in general. One of the key robustness requirements is the connectivity between each pair of nodes through a sufficiently short path, which makes a network cluster more robust with respect to potential disruptions such as man-made or natural disasters. If one can reinforce the network components against future threats, the goal is to determine optimal reinforcements that would yield a cluster with minimum risk of disruptions. We propose a risk-averse model where clusters represents a R-robust 2-club, which by definition is a subgraph with at least R node/edge disjoint paths connecting each pair of nodes, where each path consists of at most 2 edges. And, develop a combinatorial branch-and-bound algorithm to compare with an equivalent mathematical programming approach on random and real-world networks.

Optimization is used in daily practice with fixed input quantities and assuming constancy of all internal and external factors that may affect the results. But the reality that surrounds us is not so straightforward and clear. It is a complex and variable system. We learn new information about it every moment and it changes and evolves constantly. If the variability is included in the optimization model, it will change from the deterministic to stochastic model. Simulation offers a way how to work with the variability of inputs and the risk that the future development will be different from the assumptions. It uses probability distributions of entering risk factors. The most frequent method is Monte Carlo simulation, which is based on the generation of large amount of scenarios of possible future developments. Crystal Ball was used as simulation software program.

This thesis deals with Canadian traveller problem (CTP), which can be defined as the shortest path problem in a stochastic environment. The overview of different CTP variants is presented in theoretical part of this thesis, as well as known solutions to these variants. In the next parts, the thesis focuses on the stochastic variation of CTP (SCTP). For this variant chosen solutions (strategies) are discussed more in depth. At the same time, the original strategies named UCTO and UCTP are presented. Further, the thesis deals with the description of a window application implemented in Java, which has been developed to validate and test the functionality of selected strategies. The final part contains experiments and comparison of selected strategies.

abstract: The mobile crowdsensing (MCS) applications leverage the user data to derive useful information by data-driven evaluation of innovative user contexts and gathering of information at a high data rate. Such access to context-rich data can potentially enable computationally intensive crowd-sourcing applications such as tracking a missing person or capturing a highlight video of an event. Using snippets and pictures captured from multiple mobile phone cameras with specific contexts can improve the data acquired in such applications. These MCS applications require efficient processing and analysis to generate results in real time. A human user, mobile device and their interactions cause a change in context on the mobile device affecting the quality contextual data that is gathered. Usage of MCS data in real-time mobile applications is challenging due to the complex inter-relationship between: a) availability of context, context is available with the mobile phones and not with the cloud, b) cost of data transfer to remote cloud servers, both in terms of communication time and energy, and c) availability of local computational resources on the mobile phone, computation may lead to rapid battery drain or increased response time. The resource-constrained mobile devices need to offload some of their computation. This thesis proposes ContextAiDe an end-end architecture for data-driven distributed applications aware of human mobile interactions using Edge computing. Edge processing supports real-time applications by reducing communication costs. The goal is to optimize the quality and the cost of acquiring the data using a) modeling and prediction of mobile user contexts, b) efficient strategies of scheduling application tasks on heterogeneous devices including multi-core devices such as GPU c) power-aware scheduling of virtual machine (VM) applications in cloud infrastructure e.g. elastic VMs. ContextAiDe middleware is integrated into the mobile application via Android API. The evaluation consists of overheads and costs analysis in the scenario of ``perpetrator tracking" application on the cloud, fog servers, and mobile devices. LifeMap data sets containing actual sensor data traces from mobile devices are used to simulate the application run for large scale evaluation. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2019

The deterministic theory of graphs and networks is used successfully in cases where no random component is needed. However in practice, a number of decision-making and conflict situations require the inclusion of a stochastic element directly into the model. The objective of this thesis is the introduction of stochastic optimization and its application on random networks. The reader will become familiar with three approaches to stochastic optimization. Namely two-stage optimization, multi-stage optimization and chance constraint optimization. Finally, the studied issue is demonstrated on a real telecommunication network example.

My PhD thesis concentrates on the field of stochastic analysis, with focus on stochastic optimization and applications in finance. It is composed of two parts: the first part studies an optimal stopping problem, and the second part studies an optimal control problem. The first topic considers a one-dimensional transient and downwards drifting diffusion process X, and detects the optimal times of a random time(denoted as ρ). In particular, we consider two classes of random times: (1) the last time when the process exits a certain level l; (2) the time when the process reaches its maximum. For each random time, we solve the optimization problem infτ E[λ(τ- ρ)+ +(1-λ)(ρ - τ)+] overall all stopping times. For the last exit time, the process should stop optimally when it runs below some fixed level k the first time, where k is the solution of an explicit defined equation. For the ultimate maximum time, the process should stop optimally when it runs below a boundary which is the maximal positive solution (if exists) of a first-order ordinary differential equation which lies below the line λs for all s > 0 . The second topic solves an optimal consumption and investment problem for a risk-averse investor who is sensitive to declines than to increases of standard living (i.e., the investor is loss averse), and the investment opportunities are constant. We use the tools of stochastic control and duality methods to solve the resulting free-boundary problem in an infinite time horizon. Briefly, the investor consumes constantly when holding a moderate amount of wealth. In bliss time, the investor increases the consumption so that the consumption-wealth ratio reaches some fixed minimum level; in gloom time, the investor decreases the consumption gradually. Moreover, high loss aversion tends to raise the consumption-wealth ratio, but cut the investment-wealth ratio overall.