PhD thesis defense to be held on November 24, 2023, at 16:00 (MS Teams)
Picture Credit: Elissaios Sarmas
Thesis title: Artificial Intelligence Models and Optimization Algorithms for Distributed Generation, Microgrid Flexibility and Buildings Energy Management
Abstract: The energy sector is currently experiencing significant and unprecedented changes due to several factors, including the pressing concern about the sustainability of the planet. The Paris Agreement of 2015, which calls for the design and implementation of sustainable, strong, and socially acceptable policies to combat climate change globally, is a clear indication of this concern. Addressing this challenge requires a concerted effort to transform and shift the energy sector from fossil fuel-based production and consumption systems to alternative energy sources such as solar, wind, hydrogen, and lithium-ion batteries, among others. This process of energy transition is complex and has multiple social, technological, and environmental implications beyond the goal of decarbonization.
This PhD thesis aims to contribute to this global goal by developing modern learning methods and mathematical optimization algorithms for the energy and building sector, with a specific focus on three broad categories (Renewable Energy Sources, Distributed Energy Resources at the microgrid level, and Energy Efficiency of Buildings). Through the development of artificial intelligence and mathematical optimization models, the thesis presents solutions that address problems in each of these categories by thoroughly analyzing the key parameters and interactions between them.
The contribution of the thesis to humanity lies in the development of an integrated library of artificial intelligence models and optimization algorithms that can significantly improve the efficiency and effectiveness of energy management, renewable energy sources, and energy efficiency of buildings. The thesis proposes an integrated methodology for the selection of models, algorithms, and input features for energy problems of different problem classes and their integration into a single methodological framework. In addition, the thesis provides a thorough analysis of the interdependence of the three main application areas, which is essential for the development of effective models and algorithms.
By providing practical validation of the proposed models and algorithms using real-world data and test cases, the thesis offers a tangible contribution to the global effort to combat climate change through energy transition. The proposed solutions can facilitate the transition to sustainable energy systems by addressing challenges related to the Decarbonization, Digitization, Decentralization, and Democratization of energy. Ultimately, this research project aims to make a meaningful and lasting impact on humanity by advancing our understanding of energy and buildings and by providing practical solutions to address the pressing challenges we face today.
Finally, in addition to the contribution to the field of energy and building sector optimization, the thesis proposes novel techniques in the application of artificial intelligence. Transfer learning, incremental learning and meta-learning are introduced as novel approaches for solving the problems related to renewable energy sources, distributed energy resources and energy efficiency of buildings. These techniques enable the development of models that can be adapted to different applications, allowing for the optimization of the energy sector in a more efficient and effective manner, paving the way for future research in this field.
Supervisor: Professor John Psarras
PhD Student: Elissaios Sarmas