Global-Decision-Focused Neural ODEs for Proactive Grid Resilience Management

2025-02-25

AI Summary

To address the resilience deficiency in power systems under extreme disasters (e.g., wildfires, hurricanes) caused by misalignment between forecasting and optimization objectives, this paper proposes the Prediction-Aware Total Optimization with Global coordination (PATOG) framework. PATOG jointly optimizes spatiotemporal outage dynamics modeling and global resilience decision-making, introducing the first decision-focused neural ordinary differential equation (ODE) model to enable end-to-end, decision-aware co-learning of outage prediction and intervention. By integrating neural differential equations, prediction–optimization joint training, spatiotemporal graph modeling, and global optimization embedding, PATOG achieves significant improvements on both synthetic and real-world grid datasets: 23.6% enhancement in prediction consistency (measured by MAE stability), 41% reduction in resource scheduling latency, and 37% shorter system recovery time. PATOG is the first framework to achieve spatiotemporally consistent unification of prediction and optimization objectives under disaster scenarios.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Search and Optimization: Learning to Search
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Extreme hazard events such as wildfires and hurricanes increasingly threaten power systems, causing widespread outages and disrupting critical services. Recently, predict-then-optimize approaches have gained traction in grid operations, where system functionality forecasts are first generated and then used as inputs for downstream decision-making. However, this two-stage method often results in a misalignment between prediction and optimization objectives, leading to suboptimal resource allocation. To address this, we propose predict-all-then-optimize-globally (PATOG), a framework that integrates outage prediction with globally optimized interventions. At its core, our global-decision-focused (GDF) neural ODE model captures outage dynamics while optimizing resilience strategies in a decision-aware manner. Unlike conventional methods, our approach ensures spatially and temporally coherent decision-making, improving both predictive accuracy and operational efficiency. Experiments on synthetic and real-world datasets demonstrate significant improvements in outage prediction consistency and grid resilience.

Problem

• Enhances power grid resilience against extreme hazards.
• Aligns prediction and optimization for better resource allocation.
• Integrates outage prediction with globally optimized interventions.

Innovation

• Global-Decision-Focused Neural ODEs
• Predict-all-then-optimize-globally framework
• Spatially and temporally coherent decision-making