Data-Driven Power System Optimal Decision Making Strategy under Wildfire Events

Presenter Information

Wanshi Hong, Lawrence Berkeley National LaboratoryFollow
Bin Wang, Lawrence Berkeley National LaboratoryFollow
Mengqi Yao, University of California, BerkeleyFollow
Duncan Callaway, UC BerkeleyFollow
Larry Dale, University of California, BerkeleyFollow
Can Huang, Lawrence Livermore National LaboratoryFollow

Location

Online

Event Website

https://hicss.hawaii.edu/

Start Date

3-1-2022 12:00 AM

End Date

7-1-2022 12:00 AM

Description

Wildfire activities are increasing in the western United States in recent years, causing escalating threats to power systems. This paper developed an optimal and data-driven decision-making framework that improves power system resilience under wildfire risks. An optimal load shedding plan is formulated based on optimal power flow analysis. To avoid power system cascading failure caused by wildfire, we added additional transmission line flow constraints based on the identification of power lines with high ignition risk. Finally, a data-driven method is developed, leveraging multiple machine learning techniques, to model the complex correlations between input wildfire scenarios and the output power management strategy with significantly reduced computational complexities. The proposed data-driven decision-making framework can reduce the safety impacts on the electricity consumers, improve power system resilience under wildfire events.