Nous nous intéressons à prévoir séquentiellement une suite arbitraire d'observations. À chaque instant, des experts nous proposent des prévisions de la prochaine observation. Nous formons alors notre prévision en mélangeant celles des experts. C'est le cadre de l'agrégation séquentielle d'experts. L'objectif est d'assurer un faible regret cumulé. En d'autres mots, nous souhaitons que notre perte cumulée ne dépasse pas trop celle du meilleur expert sur le long terme. Nous cherchons des garanties très robustes~: aucune hypothèse stochastique sur la suite d'observations à prévoir n'est faite. Celle-ci est supposée arbitraire et nous souhaitons des garanties qui soient vérifiées quoi qu'il arrive. Un premier objectif de ce travail est l'amélioration de la performance des prévisions. Plusieurs possibilités sont proposées. Un exemple est la création d'algorithmes adaptatifs qui cherchent à s'adapter automatiquement à la difficulté de la suite à prévoir. Un autre repose sur la création de nouveaux experts à inclure au mélange pour apporter de la diversité dans l'ensemble d'experts. Un deuxième objectif de la thèse est d'assortir les prévisions d'une mesure d'incertitude, voire de prévoir des lois. Les applications pratiques sont nombreuses. En effet, très peu d'hypothèses sont faites sur les données. Le côté séquentiel permet entre autres de traiter de grands ensembles de données. Nous considérons dans cette thèse divers jeux de données du monde de l'énergie (consommation électrique, prix de l'électricité,...) pour montrer l'universalité de l'approche. / We are interested in online forecasting of an arbitrary sequence of observations. At each time step, some experts provide predictions of the next observation. Then, we form our prediction by combining the expert forecasts. This is the setting of online robust aggregation of experts. The goal is to ensure a small cumulative regret. In other words, we want that our cumulative loss does not exceed too much the one of the best expert. We are looking for worst-case guarantees: no stochastic assumption on the data to be predicted is made. The sequence of observations is arbitrary. A first objective of this work is to improve the prediction accuracy. We investigate several possibilities. An example is to design fully automatic procedures that can exploit simplicity of the data whenever it is present. Another example relies on working on the expert set so as to improve its diversity. A second objective of this work is to produce probabilistic predictions. We are interested in coupling the point prediction with a measure of uncertainty (i.e., interval forecasts,…). The real world applications of the above setting are multiple. Indeed, very few assumptions are made on the data. Besides, online learning that deals with data sequentially is crucial to process big data sets in real time. In this thesis, we carry out for EDF several empirical studies of energy data sets and we achieve good forecasting performance.

Orientador: Elnatan Chagas Ferreira / Tese (doutorado) - Universidade Estadual de Campinas, Faculdade de Engenharia Elétrica e de Computação / Made available in DSpace on 2018-08-26T21:41:11Z (GMT). No. of bitstreams: 1 Duarte_LuisFernandoCaparroz_D.pdf: 2460153 bytes, checksum: ff832dc604aa88b406f8f597eacaa933 (MD5) Previous issue date: 2015 / Resumo: A preocupação com o futuro dos recursos energéticos leva os pesquisadores a procurar formas mais eficientes de gerenciar estes recursos. Neste trabalho apresenta-se a implementação de um sistema não invasivo de monitoramento de consumo de energia elétrica baseado em conjuntos nebulosos, capaz de informar ao cliente, em período próximo de tempo real, sobre o consumo energético das principais cargas elétricas na unidade consumidora. Este trabalho parte da premissa de que a necessidade de se conhecer com detalhes como se dá o uso da energia elétrica em uma residência é primordial para se alcançar edificações com alto grau de eficiência energética. O conhecimento preciso de como a energia elétrica está sendo consumida é fundamental para que as pessoas entendam como é possível reduzir o desperdício e maximizar economia sem que haja prejuízo na qualidade de vida. O sistema proposto baseia-se em medição centralizada de energia elétrica. Para processar o consumo dos aparelhos em uso, utiliza a medição da tensão, da potência ativa e das cinco primeiras harmônicas ímpares da corrente elétrica como notação das assinaturas de cargas, medidas em nível de circuito, logo após os disjuntores de uma caixa de distribuição elétrica. A desagregação do consumo por aparelho é resolvida empregando técnicas de conjuntos nebulosos. Como o sistema possui conexão com a Internet, além de informar seus usuários sobre o consumo energético de suas cargas elétricas, ele também pode ser usado como ferramenta para enviar informações pertinentes às empresas geradoras e distribuidoras de energia elétrica, favorecendo o desenvolvimento de redes inteligentes de abastecimento energético / Abstract: Concerns about the future of energy resources lead the researchers to pursue more efficient ways to manage these resources. This work presents the implementation of a non-invasive power consumption monitoring system based on fuzzy sets, able to inform the customer, in a period close to real time, about the energy consumption of the main electrical appliances in their consumer units. This work relies on the premise that knowing in detail how do we make use of electricity in a residence is essential to achieve buildings with high degree of energy efficiency. The precise knowledge on how the electricity is being consumed is fundamental to understand how it is possible to reduce wastage and maximize savings without loss of quality of life. The proposed system is based on centralized measurement of electrical energy. To process the consumption of the appliances in use, it makes use of measurements of voltage, active power and the first five odd harmonics of the electrical current to constitute the load signatures, sensed at circuit level, just after the breakers at an electrical distribution box. The power consumption disaggregation is solved using fuzzy sets techniques. Since the system has Internet connection, in addition to informing its users about the energy consumption of their electrical loads, it is also possible to send relevant information to electric utilities, supporting the Smart Grid development / Doutorado / Eletrônica, Microeletrônica e Optoeletrônica / Doutor em Engenharia Elétrica

Submitted by Renata Lopes (renatasil82@gmail.com) on 2015-12-07T10:24:44Z No. of bitstreams: 1 tuanneferreiradias.pdf: 1099690 bytes, checksum: 05da37de69cda810521f3f31c67e3bbd (MD5) / Approved for entry into archive by Adriana Oliveira (adriana.oliveira@ufjf.edu.br) on 2015-12-07T21:38:15Z (GMT) No. of bitstreams: 1 tuanneferreiradias.pdf: 1099690 bytes, checksum: 05da37de69cda810521f3f31c67e3bbd (MD5) / Made available in DSpace on 2015-12-07T21:38:15Z (GMT). No. of bitstreams: 1 tuanneferreiradias.pdf: 1099690 bytes, checksum: 05da37de69cda810521f3f31c67e3bbd (MD5) Previous issue date: 2015-04-24 / CAPES - Coordenação de Aperfeiçoamento de Pessoal de Nível Superior / O presente estudo tem como objetivo obter estimativas de elasticidades-preço, renda e relativas a outras variáveis para a demanda domiciliar de eletricidade segundo 20 grupos da população brasileira em 2008/2009. Os grupos são definidos segundo classes socioeconômicas e regiões macroeconômicas. Tal objetivo é motivado pelo fato de haver poucos estudos na literatura para a demanda domiciliar de eletricidade usando microdados de forma tão pouco desagregada. Para tanto, são usados microdados domiciliares da POF de 2008/2009 do IBGE. A POF 2002/2003 também é usada com o objetivo de comparação entre os dados na análise descritiva, mas não de estimação. Sendo a POF uma amostra de dados complexos, para obter resultados coerentes, considera-se o plano amostral nas estimações e também os pesos de expansão, o que leva ao uso de mínimos quadrados ponderados. Os resultados sugerem que as regiões mais sensíveis ao aumento do preço da energia são Centro-Oeste e Sul e as menos sensíveis Norte e Nordeste, ou seja, o consumo para essas últimas regiões pouco se alterariam com o aumento do preço. Já um aumento na renda domiciliar, as regiões Norte, Nordeste e Sudeste aumentariam mais o consumo de energia elétrica do que as regiões Centro-Oeste e Sul. / This study develops estimates of price and income elasticities, and with regard to other variables, for household electricity demand across 20 groups of the Brazilian population in 2008/2009. The groups are set according to socioeconomic classes and macroeconomic regions. The motivation is the fact that there are few studies in the literature for household electricity demand using microdata so as little broken. Thus, we used household microdata from IBGE’s survey of family budgets POF 2008/2009. The POF 2002/2003 are also used for comparison purposes within the descriptive data analysis, but are not used for estimation. As the POF a complex data survey, in order to obtain consistent results the sampling plan is considered in the estimates including the expansion weights, which led to the use of weighted least squares. The results suggest that regions which are most sensitive to rising energy prices are the Midwest and South, while the North and Northeast are less sensitive, ie the consumption for the latter regions would change little with the price increase. Yet an increase in household income, the North, Northeast and Southeast further increase the power consumption than the Midwest and South.

Diploma thesis deals with the problem of electric energy utilization during the construction realization. The theoretical part elaborates the topics of construction project management and fundamental methods of construction price determination. Moreover, it explains the topic of construction site equipment and describes the employment and the expensiveness of the electric energy on the construction site. The practical part of the diploma thesis shows confrontation of the theoretical background with practical use. A detailed description of the total costs of electric energy was calculated based on the study of model building. The main objective of the thesis was to characterize the influence of the electricity price variation on the total expense of the construction and proposes the optimization of the electric energy cost.

Ma thèse se compose de deux parties indépendantes. La première traite decrypto-économie et la seconde d’économétrie théorique. Dans le premier chapitre, je présente un modèle qui prédit la puissance de calcul totale déployée par les mineurs en utilisant le taux de change bitcoin / dollar. Le deuxième chapitre s’appuie sur une version simplifiée du précédent modèle pour faire le constat de l’inefficacité du protocole Bitcoin actuel et proposer un moyen simple de réduire la consommation d’électricité engendrée par cette cryptomonnaie. Le troisième chapitre explique comment identifier et estimer les bornes exactes de la région d’identification de l’effet marginal moyen dans un modèle logit avec effets fixes sur données de panel. / My dissertation concatenates two independent parts. The first one dealswith crypto-economics whereas the second one is about theoretical econometrics. In the first chapter, I present a model which predicts bitcoin miners’ total computing power using the bitcoin / dollar exchange rate. The second chapter builds on a simplified version of the preceeding model to show to which extent the current Bitcoin protocol is inefficient and suggest a simple solution to lower the cryptocurrency’s electricity consumption. The third chapter explains how to identify and estimate the sharp bounds of the average marginal effect’s identification region in a fixed effects logit model with panel data.

The increasing volume of smart electricity meter readings presents a challenge forelectricity providing companies in accurately validating and correcting the associated data. This thesis attempts to find a possible solution through the application ofunsupervised machine learning for detection of anomalous readings. Through thisapplication there is a possibility of reducing the amount of manual labor that is required each month to find which meters are necessary to investigate. A solution tothis problem could prove beneficial for both the companies and their customers. Itcould increase abnormalities detected and resolve any issues before having a significant impact. Two possible algorithms to detect anomalies within these meters areinvestigated. These algorithms are the Isolation Forest and a Autoencoder, wherethe autoencoder showed results within the expectations. The results shows a greatreduction of the manual labor that is required up to 96%.

En stark motivering till omställning från icke-förnybar till förnybar energi är ökningen av elförbrukning i världen, med målet att minska klimatpåverkan. Denna elförbrukning förväntas stiga från dagens 140 TWh till 190 TWh år 2045 i Sverige. I och med detta etableras nu konceptet Co-Living på den svenska bostadsmarknaden vilket innebär att människor delar på hemmet. I denna studie undersöktes en testbädd på KTH Live-In Lab där olika bostadskonfigurationer har granskats, där den första konfigurationen var singelboenden och den sista är Co-Living. Syftet med studien var att undersöka de boendes beteendeförändringar med avseende på resursförbrukning i hemmet och undersöka om det gick att se en skillnad mellan konfigurationerna. Resurserna som undersöktes var el och vatten. Studien utfördes genom att undersöka data för el- och varmvattenförbrukning samt aktiveringar och användningstid för dusch och tvättställ. Mätperioderna var på tre månader och innefattade fyra boenden i varje konfiguration. Resultatet visar en ökning i Co-Living, av elförbrukning per person på cirka 12,5% under 2021. Vattenförbrukningen per person visade en ökning på ca 56%. Det gick även att se att antalet aktiveringar, per person, för dusch och tvättställ steg med 41%. Sist i resultatet steg användningstiden, per person, för dusch och tvättställ med 132%. Slutsatsen av studien visar en märkbar ökning av resursförbrukning hos boenden i det studerade Co-Living jämfört med singelboenden. Men det finns en osäkerhet om denna förändring var på grund av Co-Living då exempelvis Covid-19 har visat sig kunna ha stor påverkan på resultatet. / One of the strong reasons for converting to renewable energy sources is the increasing amount of resource usage, with the goal to reduce the climate impact of it. The electricity usage is expected to rise from today’s 140 TWh to 190 TWh until year 2045 in Sweden. With this, the concept of Co-Living is now being established on the housing market. The concept is based on people sharing an accommodation. In this study, obtained data from a testbed called Live-In Lab was investigated. The data is obtained from different building configurations of accommodations, where the first one is single households and the last one is Co-Living. The purpose of this study was to examine the resident’s behavioral changes that affected the resource usage, to see if there were differences between the different configurations. The examined resources were electricity and water. The study was conducted by examining data for electricity and water usage, also usage time and activations for shower and basin. Data was collected for periods of three months and there were four tenants in each configuration. The results show that the usage, per person, of electricity increased by 12,5%, and water by 56% in 2021. Noticeable was also that the number of activations, per person, for shower and basin increased by 41%. Lastly the usage time for shower and basin increased by 132% per person. The conclusion for this study is that there is a noticeable increase of resource usage in the investigated Co-Living, comparing with single households. But there is a great uncertainty regarding Co-Living being the main reason for this difference, as Covid-19, for example, also has had a strong impact on resource usage.

År 2015 fastställde FN 17 globala mål för hållbar utveckling. Delmål 7.2, en del av det sjunde målet om hållbar energi för alla, anger att den globala andelen förnybar energi behöver öka för att minska effekterna av fossila bränslen. Förnybara energikällor framstår som avgörande för framtiden, med människor som viktiga aktörer. Privata fastighetsägare som överväger att investera i solceller möter utmaningar när det gäller att samla in och förstå information om ämnet. Investering i solceller är fördelaktigt ur ett miljöperspektiv,men det finns ytterligare anledningar för en investerare att överväga solceller, särskilt den ekonomiska aspekten. En investerare kan spara pengar genom att köpa mindre el från det allmänna nätet och samtidigt sälja överskottet av egenproducerad el.Examensarbete har tre mål: 1. redogöra för vilka typer av solceller som finns på marknaden, 2. fastställa hur faktorer som takriktning, taklutning och solinstrålning påverkar den årliga elproduktionen och 3. undersöka om elpriserna i södra och norra Sverige har en betydande inverkan på den årliga besparingen, försäljningen av överskottsel samt återbetalningstiden/lönsamheten. En kalkyl har utförts för en fastighet för att beräkna hur mycket el solcellerna kan producera årligen.Med hänsyn till den tid som var tillgänglig för genomförandet av arbetet, var vissa begränsningar nödvändiga. Solcellsteknik är ett omfattande ämnesområde, vilket resulterade i betydande begränsningar under forskningsprocessen. Ytterligare begränsningar inkluderade specifika investeringskostnader, såsom kostnader för profilskenor, skruvar för att säkra solpanelerna, och arbetskraftskostnader för solcellsinstallatörer.Arbetet grundades på beräkningar, platsbesök och intervju med en privat fastighetsägare. Beräkningarna grundades på olika antaganden och statistiska genomsnittsvärden, med hänsyn till potentiell vinst och solcellernas effektiva elproduktion i olika geografiska områden och för olika solcellsfabrikat. Fastighetsägaren deltog i en strukturerad intervju som genomfördes på ett hus i Vetlanda där platsbesöket ägde rum. Huset har ett sadeltakmed en area på 140 m2, som lutar mot öster och väster. Två telefonintervjuergenomfördes med två elbolag i Sverige. Ett av bolagen, där fastighetsägaren är kund, bedriver sin verksamhet i södra Sverige, medan det andra elbolaget, där huset omplacerades som en del av undersökningen, bedriver sin verksamhet i norra Sverige. De metoder som valdes för arbetet var relevanta för att uppnå målen med arbetet. Två olika scenarier undersöktes. Det första scenariot baserades på en relativt låg elförbrukning på 4500 kWh/år, medan det andra scenariot baserades på ett hushåll med en relativt hög elförbrukning på 20 000 kWh/år. Utöver solinstrålning och elpriser, undersöktes dessa två scenarier för att jämföra olika elförbrukningsscenarier och analysera hur de påverkar potentiella besparingar och vinster genom försäljning av överskottsel. Elprisernas fluktuationer och mängden solinstrålning på fastighetens plats är betydande faktorer som starkt påverkar det slutgiltiga resultatet. En strategisk investering i urvalet av lämpliga solcellspaneler och växelriktare från början kan visa sig vara effektiv för att påverka återbetalningstiden och lönsamheten för den totala investeringen i efterhand. / In 2015, the United Nations outlined 17 global goals for sustainable development, with Target 7.2 falling under the seventh goal, focusing on sustainable energy for all. This target emphasizes the necessity of increasing the global share of renewable energy to mitigate the impacts of fossil fuel consumption. Renewable energy sources are deemed essential for the future, with individuals playing a crucial role in this transition. Private property owners investing in solar cells face various challenges in gathering and comprehending information on the subject. While the environmental benefits of solar cell investments are evident, there are additional reasons for investors to consider this solution, particularly from a financial standpoint. By reducing reliance on public grid electricity and selling excess self-produced electricity, investors can both save money and contribute to sustainable practices.The thesis sets three objectives: 1. to delineate the available types of solar panels in the market, 2. to assess how factors such as roof direction, slope, and solar radiation impact annual electricity production, and 3. to explore the influence of electricity prices in southern and northern Sweden on annual savings, surplus electricity sale, and the payback period/profitability. Calculations have been conducted to estimate the annual electricity production potential of solar cells for a specific property.Considering the time constraints, certain limitations were imposed, mainly due to the vastness of solar cell technology. These limitations encompassed specific investment costs such as profile rails, screws for securing solar panels, and labour costs for installers.The study relied on calculations, site visits, and interviews with a private property owner, incorporating various assumptions and statistical averages related to potential profit and efficient electricity production in different geographical areas and from different solar cell manufacturers.Two scenarios were explored: the first, based on a relatively low electricity consumption of 4,500 kWh/year, and the second, assuming a higher consumption of 20,000 kWh/year. Beyond solar radiation and electricity prices, these scenarios were investigated to compare different consumption levels and analyse their impact on potential savings and profits through excess electricity sales.Fluctuations in electricity prices and solar radiation at the property's location significantly influence the final results. A strategic choice of solar panels and inverters from the outset can effectively impact the payback period and overall profitability of the investment.

Introduction: The global demand for renewable energy is expected to grow annually due to rising electricity consumption and economic and regulatory incentives. Efficient consumption and management of electricity consumption can support this trend (shifting from fossil resources to renewable resources) and benefit companies economically by reducing peak loads, reducing subscription costs, and protecting companies' power systems and networks. For effective and efficient electricity management that can help reduce peak electricity demand, resulting in lower demand charges and further reducing operational costs, it is necessary to understand the amount of electricity consumption and its influencing factors. Seasonal and daily variations and social behaviors influence changes in electricity consumption. Electric load variation management is essential for electricity consumers to control costs related to maximum load capacity and building electrical network equipment protection. This research presents an assessment method for describing daily electric load variations. It is applied to electricity consumption systems, but the technique is generic and can be applied to all activities where daily variations occur.Purpose: This thesis aims to show when and why peak load occurs in an office building and also to provide methods to improve the efficiency of electricity consumption during peak load.Method: This method implemented parameters like temperature, electricity consumption, and hours. It assessed a year's hourly electricity consumption in an office building to understand how load changes daily, weekly, and monthly. The case study was the AirSon office building. Data-driven from the efergy online portal was based on the hourly consumption every year.Results: The results show that the outdoor temperature, working hours, and consumption behavior can affect the overall electricity consumption and peak loads. Efficient building strategies are crucial for reducing peak loads by smartly controlling indoor temperature and managing electricity demand. The analysis offers insights into office building electricity consumption patterns and recommends strategies such as prioritizing car charging, optimizing the heat pump's operation of the HVAC system, improving the electricity management system, and using energy storage systems to reduce peak load.Discussion: The findings from this analysis carry implications for electricity consumption. It can provide some insights for reducing electricity consumption and enhancing efficiency in office buildings and similar facilities and also prioritization and rescheduling car chargers. In conclusion, it has interpreted and discussed the various electricity-saving strategies and their potential impact on electricity management.

The aim of this thesis is to describe the course of consumption of electricity. One of the feasible ways of description is statistical analysis, which enables to calculate statistically significant factors and their combinations that contributed to the course of consumption of electricity. These factors may be used for modulation of future electricity consumption, and therefore also for long-time prediction. The second chapter discusses the expansion of electricity usage in the Czech Republic from the turn of the 19th century until nowadays. The chapter describes the development of transmission system, development of electricity consumption in the sectors of the national economy, possibilities of using electric energy, and overall balance of electricity and sources of energy. The third chapter presents an overview of usage and consumption of electrical energy in the neighboring European countries as well as in the most interesting countries of the World. The fourth chapter contains statistical analysis. The first part of the chapter details a list of the analyzed quantities of individual consumptions, of the investigated factors, and the analyzed countries. Further, the chapter explains the method of statistical analysis via using simple and multiple regression and its subsequent application and evaluation for the Czech Republic and the others European countries. According to the result, it is GDP that has the main impact on the trends in the course of consumption of electricity. However, it is necessary to consider other factors influencing the consumption of electricity, too, and not rely solely on GDP in terms of long-time prediction of electricity consumption. The fifth chapter discusses the reduction of electricity consumption and savings, which can be both political and macroeconomic result. Quantification of savings is not always unequivocal and any cost-saving actions can lead to an increase in electricity consumption, which can, in effect, be bigger than any possible savings.