Identifying Deterministic and Stochastic Dynamics of Complex Energy Systems

• Applicant: Professor Dr. Benjamin Schäfer
• Subject Area: Electrical Energy Systems, Power Management, Power Electronics, Electrical Machines and Drives ; Statistical Physics, Nonlinear Dynamics, Complex Systems, Soft and Fluid Matter, Biological Physics
• Term: since 2024
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 556503410

Complex systems surround us and an increasing amount of data becomes available to analyse them. Understanding these complex systems is extremely important to operate them stably and efficiently. Let's consider the electrical energy system or power system as an example. Replacing fossil-fueled plants with renewable generators and purely passive components by power electronics increases the complexity of the power system. Further, the interplay between the increasing number of devices makes the overall dynamics less transparent and harder to estimate using detailed models of all components (bottom-up). Hence, top-down approaches for system identification via data-driven tools will play a central role in processing the increasing amounts of operational data. Simultaneously, the complex system's community has been developing inference and identification tools for both deterministic and stochastic systems. However, these are often only demonstrated on synthetic data. Within the proposed project, I will prepare an empirical data set for system identification challenges, motivated by real-world energy systems, ranging from large-scale systems to small-scale and experimental settings, such as individual houses or battery systems. On these data sets, I will develop and refine methods to infer the dynamical and stochastic equations of motion. Complementing traditional approaches, I will utilize artificial intelligence models, such as physics-informed neural networks to carry out this (deep) symbolic regression task. Overall, the project will develop inference and identification tools directly. At the same time, it will point the complex systems community to the frontiers of where and which methods are still needed. The energy system serves hereby as a very relevant example but the methodology should be generally applicable.

• DFG Programme: Research Grants