Skip to main content

Trustworthy Machine Learning for Power System Planning and Operation

T. Li, D. Deka, T. Zheng, X. Luo, M. Hong, B. Donon

  • PES
    Members: Free
    IEEE Members: $25.00
    Non-members: $40.00
    Pages/Slides: 93
Panel 20 Jul 2023

There has been an explosion of machine learning applications for power systems in recent years. However, for most of the power system industry, neural networks and other machine learning approaches remain a black box. It is, therefore, extremely difficult for utilities and power system operators to trust and adopt such tools in practice, despite the advantages they can offer. This session aims to address this challenge in two ways. First, outline the barriers the power system industry faces for the application of machine learning in power system planning and operation. Second, discuss trustworthy machine learning methods, which can provide rigorous guarantees about how such methods behave; trustworthy and explainable machine learning can build the missing trust in the power systems industry, and unlock a series of opportunities for power system applications. The goal of this session is to discuss the state-of-the-art, inform the industry, and inspire the research community to work on the development and such tools. Presentations in this panel session: - Enhancing Responsible AI for Scalable and Adaptive EV Charging (23PESGM4235) - Certifiable and Trustworthy fault localization in power grids (23PESGM4236) - Machine Learning in Grid Operation (23PESGM4238) - Increasing AI robustness to topological variations (23PESGM4240)

Pascal Van Hentenryck, Spyros Chatzivasileiadis
Primary Committee:
Power System Operation, Planning, and Economics (PSOPE)
Sponsor Committees:
Bulk Power System Operations Subcommittee