AI for Distributed Energy Systems - Jochen Cremer (TU Delft)

Conference Website: https://saiconference.com/IntelliSys

Abstract: The energy transition has to happen imminently towards net-zero. There, the distributed paradigm toward more flexible operation of the electricity system may find its new foundation in the active real-time self-management of the many distributed energy resources. There, interestingly, the recent progress of developing AI-based algorithms for distributed self-organized agents part of the system can consider some system stability constraints, and manage network congestion through their energy market participation. However, reinforced AI-based control may lack verifiable guarantees, transparency, and privacy, and can have inherent inaccuracies. In response, the system operators finally responsible for the reliable operation of the grid need advanced state estimation techniques using collected measurements so they can prevent instabilities. Processing these collected measurements is unfortunately not straightforward as the amount of data is increasing exponentially which is why advanced data processing techniques are currently in development. There, also for system operators, AI-based algorithms are very promising in the estimation of system states, the detection of active devices without intrusion, and the forecasts of some state variables such as demand or distributed renewable power injection. In this context, this keynote address will put forward a vision for developing tools for the energy transition with AI methods such as multi-agent reinforcement learning and deep learning.

Dr. Jochen Cremer is the Co-Director of the TU Delft AI Energy Lab and an Assistant Professor at the Faculty of Electrical Engineering, Mathematics, and Computer Science. He is also Research Associate at Imperial College London. Jochen’s research focuses on applying machine learning and data analytics to energy systems operation and control. The TU Delft AI Energy lab team develops new computational methods for system operation and control that combine statistical machine learning and mathematical optimisation.