Physics-informed Machine Learning for Virtual Inertia Provision from Distribution Power Systems

Simon Stock

• Publisher: Shaker
• Year: 2025
• eBook ISBN: 9783844098600
• Edition: 0

This thesis develops a framework that enables the provision of Virtual Inertia from power distribution systems. In this way, distributed renewables energies can be utilized to support the overall system frequency. Physics-informed Machine Learning techniques are developed and applied inside this framework. Namely, the Bayesian Physics-informed Neural Network and the Physics-informed Actor Critic.

Table of contents

• Introduction
• Frequency Dynamics of Power Systems
• Physics-informed Machine Learning
• Inertia Support Framework
• Estimation of System Inertia
• Coordination of Virtual Inertia Provision
• Discussion
• Conclusions and outlook
• Appendix
• Bibliography