Improving energy autonomy of positive energy districts using multi-agent deep reinforcement learning
マルチエージェント深層強化学習を用いたポジティブエネルギーディストリクトのエネルギー自律性向上 (AI 翻訳)
Šribar, Jernej, Mohorcic, Mihael, Čampa, Andrej
🤖 gxceed AI 要約
日本語
本論文は、V2G対応EVと共有ESSを協調制御するマルチエージェント深層強化学習フレームワークCoMAD V2Gを提案。実データに基づくシミュレーションで検証し、外部系統依存を削減し、自家消費を最大化。最大25%の電気代削減を達成し、プライバシー保護と拡張性を両立した持続可能なエネルギー管理戦略を示す。
English
This paper proposes CoMAD V2G, a multi-agent deep reinforcement learning framework for coordinated management of V2G-enabled EVs and shared energy storage in Positive Energy Districts. Validated with real-world datasets, it reduces grid reliance, increases local renewable utilization, and lowers household electricity costs by up to 25%, offering a scalable and privacy-preserving solution for smart energy management.
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、ポジティブエネルギーディストリクト（PED）やスマートコミュニティの実現に向けて、再エネとEVの統合制御が重要課題。本手法はプライバシー保護と経済性を両立し、日本のVPP構想や地域エネルギー自律化に示唆を与える。
In the global GX context
Globally, Positive Energy Districts are central to the EU's energy transition. This work provides a scalable, privacy-preserving MARL solution for integrating V2G and storage, reducing grid dependence and costs. Relevant to ISSB's energy transition and TCFD's climate risk management for utilities.
👥 読者別の含意
🔬研究者:Novel multi-agent RL framework for V2G and storage coordination in PEDs with real-world validation, advancing scalable energy management.
🏢実務担当者:Demonstrates up to 25% cost savings and improved self-consumption, applicable to district energy operators or microgrid managers.
🏛政策担当者:Supports policies for V2G integration and district-level energy autonomy, aligning with EU PED framework and similar initiatives.
📄 Abstract（原文）
This paper presents CoMAD V2G , a novel multi-agent deep reinforcement learning (MARL) framework designed to improve the energy autonomy of Positive Energy Districts (PEDs) through the coordinated management of Vehicle-to-Grid (V2G)-enabled electric vehicles and a shared Energy Storage System (ESS). The proposed approach addresses the challenges posed by fluctuating renewable energy generation, dynamic electricity prices and the variable availability of electric vehicles while preserving user comfort through non-intrusive control. The framework is validated using real-world datasets covering household electricity consumption, photovoltaic generation, EV mobility patterns and market electricity prices within a realistic simulation environment. Its performance is benchmarked against conventional charging strategies and an alternative reinforcement learning approach, demonstrating superior energy management across multiple scenarios. Results show that CoMAD V2G substantially improves community energy autonomy by reducing reliance on the external grid while increasing the effective use of locally generated renewable energy. The proposed solution also lowers household electricity costs by up to 25%, providing a scalable, privacy-preserving and economically sustainable strategy for intelligent energy management in future Positive Energy Districts.
🔗 Provenance — このレコードを発見したソース
- Zenodo https://zenodo.org/records/21455717first seen 2026-07-21 04:17:28
🔔 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。