abstract: The electric power system is monitored via an extensive network of sensors in tandem with data processing algorithms, i.e., an intelligent cyber layer, that enables continual observation and control of the physical system to ensure reliable operations. This data collection and processing system is vulnerable to cyber-attacks that impact the system operation status and lead to serious physical consequences, including systematic problems and failures. This dissertation studies the physical consequences of unobservable false data injection (FDI) attacks wherein the attacker maliciously changes supervisory control and data acquisition (SCADA) or phasor measurement unit (PMU) measurements, on the electric power system. In this context, the dissertation is divided into three parts, in which the first two parts focus on FDI attacks on SCADA and the last part focuses on FDI attacks on PMUs. The first part studies the physical consequences of FDI attacks on SCADA measurements designed with limited system information. The attacker is assumed to have perfect knowledge inside a sub-network of the entire system. Two classes of attacks with different assumptions on the attacker's knowledge outside of the sub-network are introduced. In particular, for the second class of attacks, the attacker is assumed to have no information outside of the attack sub-network, but can perform multiple linear regression to learn the relationship between the external network and the attack sub-network with historical data. To determine the worst possible consequences of both classes of attacks, a bi-level optimization problem wherein the first level models the attacker's goal and the second level models the system response is introduced. The second part of the dissertation concentrates on analyzing the vulnerability of systems to FDI attacks from the perspective of the system. To this end, an off-line vulnerability analysis framework is proposed to identify the subsets of the test system that are more prone to FDI attacks. The third part studies the vulnerability of PMUs to FDI attacks. Two classes of more sophisticated FDI attacks that capture the temporal correlation of PMU data are introduced. Such attacks are designed with a convex optimization problem and can always bypass both the bad data detector and the low-rank decomposition (LD) detector. / Dissertation/Thesis / Doctoral Dissertation Electrical Engineering 2017

Cette thèse concerne à la gestion globale à base de modèle des flux énergétiques dans le bâtiment. L'objectif des systèmes de gestion de l’énergie proposés est d’aider les occupants à gérer leur système bâtiment en planifiant la consommation/production des différents équipements présents en fonction des tarifs, de la disponibilité de l'énergie et des usages de l'occupant. L’objectif de cette thèse est de proposer une approche de résolution multi-solveurs s'appuyant sur le paradigme multi-agent.Pour y parvenir, nous avons développé, dans un premier temps, un service multi-phase, qui représente plusieurs phases flexibles avec des niveaux de consommation différents, offrant un niveau de précision supérieure pour des équipements comme une machine à laver. Dans un second temps, notre travail a consisté à proposer une nouvelle approche d’optimisation combinant différents solveurs embarqués dans des agents logiciels. Le résultat est une approche d’optimisation hybride à base d’agents, s’appuyant sur des algorithmes PLNE et des méta-heuristiques. / This PhD focuses on the global energy management based on energetical flux models ofbuildings. The objective of the energy management system is to help the inhabitants to manage theirbuilding, by scheduling the consumption/production of the different appliances, taking into accountenergy costs, availability et inhabitants’ preferences. The PhD objective is to propose a resolution approach with several solvers inspired by the multi agent systems.A multi-phasis service representing several phasis has been developped. Each phasis is defined by its own consumption level. This type of service presents a better precision than singlephasis approaches to model some appliances. A new optimization approach has been developped. Itcombines different solvers embedded into software agents. The results is an hybrid approach forthe optimization based on the multi-agent system, using MILP algorithms and meta-heuristics.

As Redes Elétricas Inteligentes podem ser entendidas como o uso intensivo de tecnologias de informação e comunicação nas redes elétricas, permitindo um fluxo bidirecional de informações e eletricidade pela rede, de forma a obter uma infraestrutura capaz de automaticamente monitorar, proteger e otimizar a operação de seus elementos. A modernização da infraestrutura elétrica no sentido das Redes Elétricas Inteligentes é inevitável e trará profundas mudanças em todos os segmentos do sistema elétrico. Embora a tecnologia necessária para essa modernização já exista a um custo razoável, ainda restam várias questões que devem ser resolvidas. Indefinições em relação aos padrões a serem adotados, regulamentações, segurança, privacidade e vários aspectos tecnológicos dificultam uma implementação coerente, adiando essa modernização. Assim, este trabalho busca apresentar esse cenário em relação às Redes Elétricas Inteligentes, investigando as tendências e situação atual. Dentre essas tendências, há um grande interesse em definir e implementar mecanismos que incentivem uma maior conscientização dos consumidores em relação ao seu uso de energia, além de uma participação ativa dos mesmos no mercado de energia. Isso exigirá a utilização de ferramentas que possibilitem a redução de custos através do uso mais eficiente da energia. Assim, é também proposto um framework para a construção e análise dessas ferramentas que buscam auxiliar os consumidores nesse cenário em formação. O framework proposto foi construído utilizando uma abordagem por Sistemas Multiagentes e possibilita a construção, simulação e análise de diversos sistemas, em diferentes cenários, com variados tipos de equipamentos, tanto reais como virtuais, sob diferentes protocolos de comunicação e com a possibilidade de uso de diversos algoritmos para a operação conjunta dos equipamentos. / Smart Grids can be understood as the intensive use of information and communication technologies over the electricity networks, allowing a bidirectional flow of information and electricity through the network. It is a system that tries to optimize the supply and demand of energy through the integration of distributed generation and renewable energy resources, and through the active participation of consumers as well as an intense trade relationship between all the segments of the electricity sector. The modernization of the electrical infrastructure towards Smart Grids is inevitable and it will bring deep changes in all segments of the electrical system. Although the necessary technology for this modernization already exists at a reasonable cost, there are still several issues to be solved. Uncertainties regarding standards to be adopted, regulations, security, privacy and many technological aspects difficult a consistent implementation and, therefore, delay this modernization. Thus, this study aims to present the Smart Grid scenario, by the investigation of its current situation and tendencies. Among these tendencies, there is a great interest to define and implement mechanisms to encourage consumer to take care about their electrical energy use and to stimulate their active participation in the energy market. This will require tools that will help them to reduce costs through a more efficient use of energy. Therefore, this work proposes also a framework for the development and analysis of these tools that help consumers at this scenario under construction. The proposed framework is built using a Multiagent System approach which allows the construction, simulation and analysis of various systems in different scenarios. In addition, it allows the use of several types of equipments, both real and virtual, under different communication protocols and with the possible use of various algorithms for a joint operation of all Smart Grid equipments.

A principal motivação para o surgimento do conceito de Smart Grid é a otimização do uso das redes de energia através da inserção de novas tecnologias de medição, automação e telecomunicações. A implementação desta complexa infra-estrutura produz ganhos em confiabilidade, eficiência e segurança operacional. Além disso, este sistema tem como principais objetivos promover a geração distribuída e a tarifa diferenciada de energia para usuários residenciais, provendo ferramentas para a participação dos consumidores no gerenciamento global do fornecimento de energia. Considerando também o uso de dispositivos de armazenamento de energia, o usuário pode optar por vender ou armazenar energia sempre que lhe for conveniente, reduzindo a sua conta de energia ou, quando a geração exceder a demanda de energia, lucrando através da venda deste excesso. Esta pesquisa propõe um Sistema Inteligente de Suporte à Decisão baseado em técnicas de aprendizado por reforço como uma solução para o problema de decisão sequencial referente ao gerenciamento de energia de uma Smart Home. Resultados obtidos mostram um ganho significativo na recompensa financeira a longo prazo através do uso de uma política obtida pela aplicação do algoritmo Q-Learning, que é um algoritmo de aprendizado por reforço on-line, e do algoritmo Fitted Q-Iteration, que utiliza uma abordagem diferenciada de aprendizado por reforço ao extrair uma política através de um lote fixo de transições adquiridas do ambiente. Os resultados mostram que a aplicação da técnica de aprendizado por reforço em lote é indicada para problemas reais, quando é necessário obter uma política de forma rápida e eficaz dispondo de uma pequena quantidade de dados para caracterização do problema estudado. / The main motivation for the emergence of the Smart Grid concept is the optimization of power grid use by inserting new measurement, automation and telecommunication technologies into it. The implementation of this complex infrastructure also produces gains in reliability, efficiency and operational safety. Besides, it has as main goals to encourage distributed power generation and to implement a differentiated power rate for residential users, providing tools for them to participate in the power grid supply management. Considering also the use of energy storage devices, the user can sell or store the power generated whenever it is convenient, reducing the electricity bill or, when the power generation exceeds the power demand, make profit by selling the surplus in the energy market. This research proposes an Intelligent Decision Support System as a solution to the sequential decision-making problem of residential energy management based on reinforcement learning techniques. Results show a significant financial gain in the long term by using a policy obtained applying the algorithm Q-Learning, which is an on-line Reinforcement Learning algorithm, and the algorithm Fitted Q-Iteration, which uses a different reinforcement learning approach called Batch Reinforcement Learning. This method extracts a policy from a fixed batch of transitions acquired from the environment. The results show that the application of Batch Reinforcement Learning techniques is suitable for real problems, when it is necessary to obtain a fast and effective policy considering a small set of data available to study and solve the proposed problem.

As Redes Elétricas Inteligentes podem ser entendidas como o uso intensivo de tecnologias de informação e comunicação nas redes elétricas, permitindo um fluxo bidirecional de informações e eletricidade pela rede, de forma a obter uma infraestrutura capaz de automaticamente monitorar, proteger e otimizar a operação de seus elementos. A modernização da infraestrutura elétrica no sentido das Redes Elétricas Inteligentes é inevitável e trará profundas mudanças em todos os segmentos do sistema elétrico. Embora a tecnologia necessária para essa modernização já exista a um custo razoável, ainda restam várias questões que devem ser resolvidas. Indefinições em relação aos padrões a serem adotados, regulamentações, segurança, privacidade e vários aspectos tecnológicos dificultam uma implementação coerente, adiando essa modernização. Assim, este trabalho busca apresentar esse cenário em relação às Redes Elétricas Inteligentes, investigando as tendências e situação atual. Dentre essas tendências, há um grande interesse em definir e implementar mecanismos que incentivem uma maior conscientização dos consumidores em relação ao seu uso de energia, além de uma participação ativa dos mesmos no mercado de energia. Isso exigirá a utilização de ferramentas que possibilitem a redução de custos através do uso mais eficiente da energia. Assim, é também proposto um framework para a construção e análise dessas ferramentas que buscam auxiliar os consumidores nesse cenário em formação. O framework proposto foi construído utilizando uma abordagem por Sistemas Multiagentes e possibilita a construção, simulação e análise de diversos sistemas, em diferentes cenários, com variados tipos de equipamentos, tanto reais como virtuais, sob diferentes protocolos de comunicação e com a possibilidade de uso de diversos algoritmos para a operação conjunta dos equipamentos. / Smart Grids can be understood as the intensive use of information and communication technologies over the electricity networks, allowing a bidirectional flow of information and electricity through the network. It is a system that tries to optimize the supply and demand of energy through the integration of distributed generation and renewable energy resources, and through the active participation of consumers as well as an intense trade relationship between all the segments of the electricity sector. The modernization of the electrical infrastructure towards Smart Grids is inevitable and it will bring deep changes in all segments of the electrical system. Although the necessary technology for this modernization already exists at a reasonable cost, there are still several issues to be solved. Uncertainties regarding standards to be adopted, regulations, security, privacy and many technological aspects difficult a consistent implementation and, therefore, delay this modernization. Thus, this study aims to present the Smart Grid scenario, by the investigation of its current situation and tendencies. Among these tendencies, there is a great interest to define and implement mechanisms to encourage consumer to take care about their electrical energy use and to stimulate their active participation in the energy market. This will require tools that will help them to reduce costs through a more efficient use of energy. Therefore, this work proposes also a framework for the development and analysis of these tools that help consumers at this scenario under construction. The proposed framework is built using a Multiagent System approach which allows the construction, simulation and analysis of various systems in different scenarios. In addition, it allows the use of several types of equipments, both real and virtual, under different communication protocols and with the possible use of various algorithms for a joint operation of all Smart Grid equipments.

For vehicles whose duty cycle is dominated by start-stop operation, fuel consumption may be significantly improved by better management of the start-stop process. Pneumatic hybrid technology represents one technology pathway to realise this goal. Vehicle kinetic energy is converted to pneumatic energy by compressing air into air tank(s) during the braking. The recovered air is reused to supply an air starter, or supply energy to the air path in order to reduce turbo-lag. This research aims to explore the concept and control of a novel pneumatic hybrid powertrain for a city bus application to identify the potential for improvements in fuel economy and drivability. In order to support the investigation of energy management, system architecture and control methodologies, two kinds of simulation models are created. Backward-facing simulation models have been built using Simulink. Forward-facing models have been developed in the GT-POWER and Simulink co-simulation. After comparison, the fully controllable hybrid braking system is chosen to realize the regenerative braking function. A number of architectures for managing a rapid energy transfer into the powertrain to reduce turbo-lag have been investigated. A city bus energy control strategy has been proposed to realize the Stop-Start Function, Boost Function, and Regenerative Braking Function as well as the normal operations. An optimisation study is conducted to identify the relationships between operating parameters and respectively fuel consumption, performance and energy usage. In conclusion, pneumatic hybrid technology can improve the city bus fuel economy by at least 6% in a typical bus driving cycle, and reduce the engine brake torque response and vehicle acceleration. Based on the findings, it can be learned that the pneumatic hybrid technology offers a clear and low-cost alternative to the electric hybrid technology in improving fuel economy and vehicle drivability.

La thèse est une contribution à la réduction de la consommation d'énergie primaire et à une meilleure utilisation des sources d'énergie renouvelables dans le cadre des systèmes de refroidissement utilisés dans le bâtiment. Après un état de l'art sur les systèmes de refroidissement, un modèle dynamique d'un système de rafraîchissement solaire à base de machine à absorption est développé et simulé. Ensuite, un facteur d'efficacité pour comparer la pertinence de ce système dans différentes régions du monde est défini. Dans la troisième partie, la notion des systèmes de refroidissement hybride -une méthode efficace contribuant à la réduction de la consommation d'énergie primaire- est présentée. Puis, les systèmes hybrides de refroidissement sont classés en catégories et sont comparés avec les systèmes de refroidissement individuels. Ensuite, un schéma permettant de sélectionner le meilleur système de refroidissement hybride dans des conditions données est proposé. Dans la dernière partie, une méthode de dimensionnement d’un système hybride à base d’énergies renouvelables est établie. Ainsi, le dimensionnement est réalisé en tenant compte de la région spécifique d’utilisation. Pour ce faire, un système de refroidissement hybride, conçu pour une maison standard, est modélisé puis simulé en utilisant le logiciel Trnsys. Finalement, et pour illustrer la méthode proposée, la problématique de dimensionnement est considérée pour deux régions différentes du globe; à savoir Marseille-France et Beyrouth-Liban. Le but est d’évaluer les performances de la méthode, à travers des données effectives, pour diverses conditions climatiques, prix des composants et tarif d'électricité. / This thesis is a contribution towards the reduction of primary energy consumption and a better use of the renewable energy sources within the cooling system for building use. After a state of the art of the cooling machines for building use, a dynamic model for a solar absorption cooling system is developed and simulated. Then, an effectiveness factor (EF) for the comparison of solar absorption chiller suitability in different locations is defined. In the third chapter, the concept of hybrid cooling system -an efficient method contributing to the reduction of primary energy consumption- is presented. Hybrid cooling systems are categorized and reviewed, with the improvement achieved compared to standalone technologies. Then, a scheme for the selection of the best hybrid cooling system for given conditions is proposed. In the last part, an optimal sizing method that defines, in a specific region, a hybrid cooling energy system, economically feasible with maximum renewable energy share is presented. Thereby, the sizing method is performed taking into account the region where it will be used. For this purpose, a hybrid cooling system, used for a standard residential house, is designed. The system is modeled and simulated using a transient system simulation program, called Trnsys. Finally, the problem of sizing is studied for different case studies; namely Marseilles-France and Beirut-Lebanon. The aim is to assess the proposed method according to diverse climatic conditions, component prices and electricity costs.

Die wechselhafte Energiebereitstellung für drahtlose Sensorknoten durch Solarzellen stellt das Energiemanagement dieser Systeme vor große Herausforderungen. Bedingt durch saisonale und kurzfristige Effekte treten kontinuierlich Schwankungen in der Eingangsleistung auf, gleichzeitig soll jedoch eine zuverlässige und konstante Systemfunktion realisiert werden. Um dies miteinander zu vereinbaren, wird ein Modell zur Beschreibung der erwarteten Eingangsleistung aufgestellt, mit welchem der planmäßige Energieverlauf bestimmt werden kann. Dieser kann wiederum mit der realen Eingangsleistung verglichen werden, um den tatsächlichen energetischen Zustand des Sensorknotens zu bestimmen. Daraus lassen sich beispielsweise Entscheidungskriterien für die Steuerung der Energieverteilung oder Betriebszustände ableiten. Im Rahmen der Arbeit werden die physikalischen Hintergründe zur Modellierung der eingehenden Sonnenenergie beschrieben, der Stand der Technik zur Modellierung aufgezeigt und ein Modell als Basis für die weiteren Untersuchungen ausgewählt. Dieses wird auf die stark limitierte Hardware von drahtlosen Sensorknoten angepasst. Die Herausforderungen liegen dabei hauptsächlich in der geringen verfügbaren Rechenleistung, wenig Datenspeicher im System und dem Ziel, möglichst wenig Energie für die Berechnung zu verbrauchen. Im Ergebnis zeigt sich, dass ein angepasstes Modell auf drahtlosen Sensorsystemen umgesetzt werden kann und trotz der starken Limitierungen lauffähig ist. Es wird eine deutliche Verbesserung in der Verteilung der Energie über den Tag ermöglicht, wodurch sich trotz wechselhafter Quelle eine konstante Systemfunktion ergibt. Nebenher wird die Zuverlässigkeit und Ausfallsicherheit erhöht und Überdimensionierungen in Energiespeicher und Solarzelle können verringert werden. Das modellbasierte Energiemanagement stellt somit einen wichtigen Baustein für eine gesicherte Energieversorgung drahtloser Sensorsysteme dar. / The volatile energy supply by solar cells for wireless sensor nodes causes vast challenges for the energy management of such systems. Conditioned by seasonal and short time effects, the incoming power continuously varies. Simultaneously a reliable and constant function of the system has to be realized. To reconcile this, a model for the expected incoming solar power has been derived, which enables the estimation of the planned energy curve. This curve can be compared with the real progression of incoming power measured in parallel, to determine the current state of energy of a sensor node. This comparison is used to derive decision criteria for the control of the energy distribution or operating conditions. Within this work, the physical backgrounds for the modelling of the incoming solar energy and the state of the art of modelling solar power are described. A model is chosen as basis for further investigations and adapted to the limited hardware of wireless sensor nodes. The main challenges are the reduced processing power, few data memory in the system and the objective to consume as few energy as possible for the calculation. The results show that an adapted model can be implemented on wireless sensor systems and that it is executable despite the heavy limitations. This enables a distinct improvement of the distribution of energy across the day, which results in a constant systems function, despite the varying incoming power. At the same time the reliability and failure safety are being improved and the oversizing of the solar cell and the storage elements can be reduced. Therefore the model based energy management is an important component for a stable power supply of wireless sensor systems.

Ce travail a trait à la gestion des flux énergétiques électriques au sein du réseau embarqué d’un véhicule électrique. Les éléments constitutifs de la chaîne électrique ont été d’abord modélisés à des fins de commande et de simulation. Il est visé ici la minimisation du stress des batteries au plomb via une hybridation avec des supercondensateurs. Deux familles de lois de commande ont été conçues et développées, à savoir des lois de type « fréquentielles » et des lois optimales de type « Linéaires Quadratiques Gaussiennes ». Un banc de test temps réel hybride a été architecturé afin de tester ces lois. Ce banc de test a pour noyau deux simulateurs temps réel (RT-LAB et dSPACE). Une partie de la chaîne de puissance est soit émulée par des sources contrôlées ou réalisée via des maquettes à échelle réduite mais à facteur de similitude respecté. Les essais sur le banc de test ont permis d’obtenir des résultats satisfaisants et encourageants qui corroborent la théorie. / This work addresses the management of electrical energy flows within the embedded network of anelectric vehicle. The electrical system components were first modeled for purposes of control synthesisand simulation. It is aimed the minimization of lead-acid batteries stress via hybridization withultracapacitors. Two families of control laws have been conceived and developed, namely afrequency-domain-based law and an optimal Linear Quadratic Gaussian law. A real-time hybrid testbench has been built in order to test these laws. This test bench has two core real-time simulators (RTLABand dSPACE). A part of the power chain has been either emulated by using suitably-controlledsources or realized by using small-scale real hardware with the similarity factor being respected. estson the testbed have yielded satisfactory and encouraging results that corroborate the theory.

No description available.