Applications of computer/network science in cyber-physical power and energy systems

Organiser

• Coventry and Warwickshire Local Network

Further Information

Presentation Abstract:

Energy systems/networks are shifting from centralized, almost deterministic generation-transmission-distribution to decentralized architectures with intermittent, uncertain distributed renewable energy sources and intelligent demand, making them complex networks. Energy informatics aims to harness computer science and AI/ML to manage these uncertainties as a self-organised complex system, as we see in the Internet. In this talk, we first review this paradigm shift and then discuss two specific applications of energy informatics: predicting outages and failure cascading using AI models and grid-friendly energy resource management. In the first application, we discuss how physics constraints and previously collected outage logs can train an AI model to predict network behavior under unseen stress testing during training or for predictive maintenance. In the second application, we discuss how distributed energy resources, such as battery storage systems, can be operated in a grid-friendly manner to alleviate congestion and avoid costly grid reinforcement, using a two-stage control framework in which a global controller derives constraints to ensure active congestion relief and a local controller optimizes resource setpoints to maximize self-consumption or revenue subject to these constraints.