Multi-Market Bidding Optimization for Co-located RES-BESS Systems Based on SAC-TCN and Flexible Arbitrage
SAC-TCNと柔軟なアービトラージに基づくRES-BESS併設システムの複数市場入札最適化 (AI 翻訳)
Shu Li, Yu Song, Zhongli Bai
🤖 gxceed AI 要約
日本語
本論文は、再生可能エネルギーと蓄電池の併設システム（RES-BESS）が複数の電力市場に参加する際の入札戦略を、強化学習（SAC）と時系列特徴抽出（TCN）を用いて最適化する枠組みを提案する。固定アービトラージの制約を撤廃し、連続行動空間で充放電の方向・量・タイミングを自律決定する。アイルランドの実データに基づくシミュレーションで、年間利益がベースライン比14%向上し、堅牢性と解釈可能性も改善した。
English
This paper proposes a flexible arbitrage framework for co-located renewable-battery systems (RES-BESS) bidding in multiple electricity markets, using Soft Actor-Critic (SAC) and Temporal Convolutional Network (TCN) to autonomously decide charging/discharging actions in a continuous space. Simulations on real Irish market data show a 14% annual profit improvement over baseline, with enhanced robustness and interpretability.
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
日本では再エネ導入拡大に伴い、蓄電池併設による系統安定化と収益性向上が課題。本手法は、容量市場や需給調整市場など複数市場への入札最適化に応用可能で、FIP制度下での再エネ事業者の収益改善や、電力市場改革に対応する実務に示唆を与える。
In the global GX context
Globally, the integration of renewables and storage is critical for grid flexibility and decarbonization. This work advances AI-driven bidding strategies for multi-market participation, relevant to market design and renewable integration policies under the energy transition.
👥 読者別の含意
🔬研究者:Provides a novel RL-based framework for multi-market bidding optimization, demonstrating the value of SAC-TCN in energy arbitrage.
🏢実務担当者:Offers a practical approach for RES-BESS operators to enhance profitability and operational flexibility in electricity markets.
🏛政策担当者:Highlights the potential of AI in enabling efficient renewable integration, informing market design and storage incentives.
📄 Abstract（原文）
Co-located renewable energy-battery energy storage systems (RES-BESS) face decision-making problems in complex time-varying environments when participating in multiple electricity markets, and traditional fixed arbitrage approaches cannot fully exploit the flexible regulation capability of BESS. To address this, this paper proposes a flexible arbitrage decision framework for multi-market bidding of co-located RES-BESS systems, which removes the predefined arbitrage time slots and frequency limits imposed by fixed arbitrage and enables the agent to autonomously determine the charging/discharging direction, power magnitude, and timing within a continuous action space. Within this framework, Soft Actor-Critic (SAC) is introduced to enhance policy exploration in the complex action space, and a Temporal Convolutional Network (TCN) is employed to extract multi-scale temporal features for better modeling of market price dynamics. Simulation results based on real Irish electricity market data show that SAC+TCN+Flex achieves an average annual profit of 798 kEUR over a 345-day evaluation set, corresponding to a 14% improvement over the baseline. On the 20-day test set, it delivers a 7.3% profit increase while exhibiting improved cross-seed robustness, more flexible temporal decision-making behavior, and enhanced policy interpretability.
🔗 Provenance — このレコードを発見したソース
- semanticscholar https://doi.org/10.1109/ccssta69471.2026.11635180first seen 2026-08-14 05:50:55 · last seen 2026-08-17 05:42:38
🔔 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。