Adequacy- and Resilience-Informed Decision-Focused Learning for High-Renewable Power Systems During Weather-Induced Operational Stress Events
気象誘発運転ストレス事象時の高再生可能エネルギー電力系統のための十分性およびレジリエンスを考慮した意思決定焦点型学習 (AI 翻訳)
Xia Y, Ma C, Wang Z, Huang Y, Huang Y, Lei (GE) S
🤖 gxceed AI 要約
日本語
本論文は、気象による運転ストレス事象下での高再生可能エネルギー電力系統の運用改善を目的とし、予測と最適化を統合した意思決定焦点型学習（DFL）フレームワークを提案する。従来の予測最適化手法と比較し、提案手法は系統の十分性と運用レジリエンスを向上させることを数値実験で示した。
English
This paper proposes a decision-focused learning framework integrating renewable generation forecast with system optimization to improve adequacy and operational resilience of high-renewable power systems under weather-induced stress events. Numerical results on IEEE test systems show superior performance over conventional predict-then-optimize approach.
Unofficial AI-generated summary based on the public title and abstract. Not an official translation.
📝 gxceed 編集解説 — Why this matters
日本のGX文脈において
日本の電力系統でも再生可能エネルギー導入拡大に伴い、気象変動への対応が重要になっている。本手法は、予測と運用最適化を連携させることで、系統運用の効率性とレジリエンス向上に貢献する可能性がある。
In the global GX context
This research addresses the global challenge of integrating high shares of renewables while maintaining grid reliability under extreme weather. The decision-focused learning paradigm offers a path to improve operational decisions beyond traditional forecasting accuracy.
👥 読者別の含意
🔬研究者:Provides a novel methodological framework for integrating forecasting and optimization in power systems under uncertainty.
🏢実務担当者:Could inform development of advanced energy management systems for grid operators facing renewable variability.
🏛政策担当者:Highlights need for regulatory frameworks encouraging deployment of resilience-oriented operational tools.
📄 Abstract（原文）
Weather-induced operational stress events are increasingly challenging power system operations by altering the supply-demand balance and pushing high-renewable power systems close to operating limits. Meanwhile, the uncertainty of renewable energy sources further increases ramping and reserve requirements, exacerbating both adequacy and operational resilience challenges during these events. In the conventional predict-then-optimize (PTO) paradigm, forecasts are trained to minimize prediction error, yet improved forecast accuracy does not necessarily lead to better operational decisions under complex system- and component-level constraints. To address this gap, this work proposes an adequacy- and resilience-informed decision-focused learning (DFL) framework that integrates renewable generation forecast with system optimization to improve decision quality considering the impact of weather-induced operational stress events. The proposed DFL approach adopts a gradient-based, end-to-end decision-focused training scheme using a surrogate loss which integrates a forecasting module and a dispatch module. This enables decision-loss feedback from the dispatch problem to be backpropagated to the forecaster. Adequacy and operational resilience are assessed using ex-post operational metrics and incorporated in-model via risk-aware chance-constraint reformulations under a multivariate Gaussian distribution. The proposed DFL-based framework is verified on the IEEE 6-node and 39-node systems. Comprehensive numerical results demonstrate that, compared with PTO, the proposed framework achieves superior adequacy and operational resilience performance under weather-induced operational stress events.
🔗 Provenance — このレコードを発見したソース
- Research Square https://doi.org/10.22541/authorea.15003119/v1first seen 2026-05-14 21:21:01
🔔 こうした論文の新着を逃したくない方は キーワードアラート に登録（無料・3キーワードまで）。
gxceed は公開メタデータに基づく研究支援データセットです。要約・翻訳・解説は AI 支援で生成されています。 最終的な解釈・検証は利用者が原典資料に基づいて行うことを前提とします。