Advancing resilient power systems through hierarchical restoration with renewable resources

Area of Science:

• Electrical Engineering
• Power Systems Engineering
• Artificial Intelligence

Background:

• Modern power systems face restoration challenges due to increased renewable energy sources (RES) and electric vehicles (EVs).
• These resources introduce significant uncertainty and flexibility, complicating post-outage recovery.
• Existing restoration methods struggle to efficiently manage these complex dynamics.

Purpose of the Study:

• To present a novel Hierarchical Modern Power System Restoration (HMPSR) model for enhanced efficiency and resilience.
• To effectively integrate renewable energy sources (solar, wind) and electric vehicles (grid-connected, mobile) into the restoration process.
• To address the uncertainties associated with RES and EV integration during power system recovery.

Main Methods:

• A two-level hierarchical architecture was developed for power system restoration.
• Graph Neural Networks (GNNs) were utilized at the upper level for fault localization and topology optimization.
• Distributionally Robust Optimization (DRO) was applied at the lower level for scenario-based dispatch planning under uncertainty.

Main Results:

• The HMPSR model demonstrated an 18.6% reduction in restoration time.
• Total restoration costs were reduced by 15.4% compared to baseline methods.
• A Grid Stability Index above 85% was maintained even under high variability conditions.

Conclusions:

• The proposed HMPSR model offers a robust and adaptive framework for modern power system restoration.
• A tightly integrated, hierarchical strategy is effective in managing complexities introduced by RES and EVs.
• The findings support the practical deployment of advanced models for resilient power grid recovery.