Pilz, Matthias (2019) Game-theoretic approaches for smart prosumer communities. (PhD thesis), Kingston University, .
Abstract
Global warming is endangering the Earth’s ecosystem. It is imperative for humanity to limit greenhouse gas emissions in order to combat rising global average temperatures. Demand-side management (DSM) schemes have widely been analysed in the context of the future smart grid. Often they are based on game-theoretic approaches to schedule the electricity consumption of its participants such that it results in small peak-to-average ratios (PAR) of the aggregated load. In order to guarantee high comfort levels for the consumer, we investigate DSM schemes on the basis of individually owned energy storage systems. The scheduling of these batteries is incentivised by a specific pricing function offered to the users. Within this thesis we cover various aspects for these type of management schemes. Firstly, we design a simple game-theoretic scheduling mechanism and analyse how the battery model, more specifically the round-trip efficiency, affects the outcome. From the simulations we find the importance of highly efficient energy storage systems for the engagement of participants. Secondly, the simple scheduling mechanism is replaced with a more advanced dynamic game, that models fine-grained control over the battery. For this novel game, we derive an analytical solution for the best response of a user, considerably speeding up the solution algorithm for the game. Furthermore, a comparison between the two games also shows the improvements in reducing the PAR of the aggregated load. Based on the augmented game, we investigate the resilience of the equilibrium solution with respect to inevitable real-world forecasting errors. One of the main findings of this thesis is reflected in the results showing the robustness of the schedules for a large number of simulated scenarios and even in the worst-case. Thirdly, we explicitly deal with the finite horizon effect that occurs due to the fixed time frame of the game mechanism. This eventually leads to a DSM system which results in a mean PAR of the aggregated load close to the optimum. Further studies show that these outcomes can be achieved due to the interaction of the households. Individual scheduling of batteries reduces the potential reduction of PAR and is especially detrimental for the robustness against forecasting errors. Fourthly, the developed model is analysed with respect to cyber-physical attacks. We develop a novel type of data-injection attack on the forecasted data and show their impact. After suggesting suitable monitoring strategies to the utility company, a game-theoretic model is employed to understand their decision making process. Finally, we investigate which battery size is optimal for such a DSM scheme. The respective experiments give insight into the different factors that determine the sizing of the battery. From the results we can infer that certain types of users only require a small scale battery system to achieve considerable gains. Overall, this thesis provides an in-depth analysis of a demand-side management scheme that can be employed by prosumers all around the world in the nearest future. Furthermore, the experiments give insights to utility companies to focus on community approaches and how they can mitigate potential cyber attacks.
Actions (Repository Editors)
![]() |
Item Control Page |