Abstract

Περίληψη: Αυτή η διπλωματική εισάγει, για πρώτη φορά στη βιβλιογραφία, την χρήση υπεργράφων για την ταχεία δημιουργία αποτελεσματικών συνασπισμών αυτόνομων πρακτόρων. Προτείνουμε ορισμένες μεθόδους σχηματισμού, που βασίζονται σε υπάρχοντες αλγορίθμους υπεργράφων, όπως οι pruning, transversal, clustering και hybrid, και εκμεταλλευόμαστε την δομή του υπεργράφου για να εντοπίσουμε πράκτορες με επιθυμητά χαρακτηριστικά. Η προσέγγισή μας επιτρέπει τον σχεδόν στιγμιαίο σχηματισμό συνασπισμών υψηλής ποιότητας, ικανοποιώντας πολλαπλές ποιοτικές απαιτήσεις. Επιπλέον, οι μέθοδοί μας κλιμακώνονται ώστε να δέχονται δεκάδες χιλιάδες πράκτορες ως είσοδο και να εμφανίζουν τα αποτελέσματα μέσα σε κλάσματα του δευτερολέπτου, με μια από αυτές να λειτουργεί με εκατομμύρια πράκτορες μέσα σε δευτερόλεπτα. Εφαρμόζουμε την προσέγγισή μας στο πρόβλημα της δημιουργίας συνασπισμών για την παροχή ρεύματος από ηλεκτρικά οχήματα προς το ηλεκτρικό δίκτυο (το λεγόμενο πρόβλημα Vehicle-to-Grid, ή V2G). Η προσέγγισή μας είναι η πρώτη που είναι σε θέση να ασχοληθεί με μεγάλης κλίμακας, και σε πραγματικό χρόνο σχηματισμό συνασπισμών για το πρόβλημα V2G, λαμβάνοντας υπ'όψιν πολλαπλά κριτήρια για τη δημιουργία των συνασπισμών ηλεκτρικών οχημάτων. Ένα προσχέδιο των ιδεών αυτών εμφανίστηκε αρχικά σε μια σύντομη δημοσίευση στο 22ο European Conference on Artificial Intelligence (ECAI-2016) και έπειτα σε μια πλήρη στο 14o European Conference on Multi-Agent Systems (EUMAS-2016). Summarization: This thesis proposes, for the first time in the literature, the use of hypergraphs for the efficient formation of effective agent coalitions. We put forward several formation methods that build on existing hypergraph pruning, transversal, clustering and hybrid algorithms, and exploit the hypergraph structure to identify agents with desirable characteristics. Our approach allows the near-instantaneous formation of high quality coalitions, adhering to multiple stated quality requirements. Moreover, our methods are shown to scale to dozens of thousands of agents within fractions of a second; with one of them scaling to even millions of agents within seconds. We apply our approach to the problem of forming coalitions to provide (electric) vehicle-to-grid (V2G) services. Ours is the first approach able to deal with large-scale, real-time coalition formation for the V2G problem, while taking multiple criteria into account for creating the electric vehicle coalitions. A sketch of these ideas appeared originally in a short paper in the 22nd European Conference on Artificial Intelligence (ECAI-2016). Afterwards, a full paper describing our work was published in the 14th European Conference on Multi-Agent Systems (EUMAS-2016).

Abstract

The Danish EDISON project has been launched to investigate how a large fleet of electric vehicles (EVs) can be integrated in a way that supports the electric grid while benefitting both the individual car owners and society as a whole through reductions in CO 2 emissions. The consortium partners include energy companies, technology suppliers and research laboratories and institutes. The aim is to perform a thorough investigation of the challenges and opportunities of EVs and then to deliver a technical platform that can be demonstrated on the Danish island of Bornholm. To reach this goal, a vast amount of research is done in various areas of EV technology by the partners. This paper will focus on the ICT-based distributed software integration, which plays a major role for the success of EDISON. Key solution technologies and standards that will accommodate communication and optimize the coordination of EVs will be described as well as the simulation work that will help to reach the goals of the project.

Abstract

International audience; Simultaneous development of photovoltaic generation and electric vehicles strengthens the solicitations on the electric power system. This paper investigates the possible synergy between these players to jointly improve the production predictability while ensuring a low carbon mobility. It stands for a step towards a quantification of its economic and environmental fallout. First a context is described for a PV-EV collaboration. Then this is gathered into an optimization problem. Grid commitment constraints, battery aging and mobility needs are here considered from the environmental point of view of equivalent primary energy. Finally, a resolution method is presented which achieve an time-efficient optimization of the power flow for each vehicle, based on upstream computed charging policies. It relies on a stochastic modeling of both vehicles availability and forecast error of the PV production. The resolution framework is the stochastic dynamic programming, coupled with on-line minimization so as to avoid the curse of dimensionality. The proposed resolution enables to compute optimal power flow for each vehicle, even among large fleets. The emphasis is here set on a versatile resolution method which could take over many detailed objective functions.

Abstract

Electric vehicles (EVs) have received wide attention due to their higher energy efficiency and lower emissions. However, the random charging and discharging behaviors of substantial numbers of EVs may lead to safety risk problems in a distribution network. Reasonable price incentives can guide EVs through orderly charging and discharging, and further provide a feasible solution to reduce the operational risk of the distribution network. Considering three typical electricity prices, EV charging/discharging load models are built. Then, a Probabilistic Load Flow (PLF) method using cumulants and Gram-Charlier series is proposed to obtain the power flow of the distribution network including massive numbers of EVs. In terms of the risk indexes of node voltage and line flow, the operational risk of the distribution network can be estimated in detail. From the simulations of an IEEE-33 bus system and an IEEE 69-bus system, the demonstrated results show that reasonable charging and discharging prices are conducive to reducing the peak-valley difference, and consequently the risks of the distribution network can be decreased to a certain extent. Document type: Article