gxceed
← 論文一覧に戻る

A Machine Learning–Based Optimal Charging Strategy for PV- Assisted Electric Vehicle Systems Incorporating Second-Life Batteries under Degradation Constraints

劣化制約下における機械学習ベースのPV併設電気自動車システムの最適充電戦略（二次電池活用） (AI 翻訳)

Ezzat A, Abdel-Khalik AS, Hamdy RA, Hamad MS

Research Squareプレプリント2026-05-20#EV・輸送Origin: Global
DOI: 10.21203/rs.3.rs-9622388/v1
原典: https://doi.org/10.21203/rs.3.rs-9622388/v1

🤖 gxceed AI 要約

日本語

本論文は、太陽光発電と電気自動車、二次電池を統合したシステムにおいて、機械学習（XGBoost）を用いた最適充電戦略を提案する。実時間の運用パラメータ（SOC、温度、PV余剰、劣化指標）に基づき充電時間クラスを予測し、離散的な充電モードで制御する。シミュレーションの結果、PV自家消費が約22%向上し、熱ストレスや高SOC曝露が低減され、バッテリー寿命延長に寄与することが示された。

English

This paper proposes a machine learning-based energy management system for PV-assisted EV and second-life battery charging, using XGBoost to predict optimal charging duration classes. The approach improves PV self-consumption by 22% while reducing thermal stress and high-SoC exposure, extending battery lifetime in residential hybrid systems.

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

日本では住宅用太陽光発電の普及と電気自動車の拡大が進んでおり、蓄電池の劣化を考慮した充電制御は、FIT終了後の自家消費最大化や系統安定化に貢献する。二次電池の活用は、EVバッテリーのリユース促進にもつながる。

In the global GX context

This study addresses the global challenge of integrating variable renewable energy and EV charging through intelligent control. The use of machine learning for degradation-aware charging is relevant for grid operators and utility companies seeking to manage distributed energy resources while extending battery life.

👥 読者別の含意

🔬研究者:The proposed ML-based discrete EMS approach provides a novel framework for optimizing EV charging under degradation constraints.

🏢実務担当者:Energy management system developers and EV charger manufacturers can leverage the OCDC prediction model to enhance battery longevity and self-consumption.

🏛政策担当者:Policymakers can use insights on second-life battery value and PV self-consumption to design incentives for residential EV charging infrastructure.

📄 Abstract（原文）

<title>Abstract</title> <p>The rapid growth of electric vehicles (EVs) and residential photovoltaic (PV) systems introduced significant challenges in managing battery charging due to renewable variability, grid constraints, and battery degradation. This paper proposes a machine learning–integrated Energy Management System (EMS) for PV-assisted smart charging of EV and second-life batteries. The system predicts the Optimal Charging Duration Class (OCDC) using an XGBoost model based on real-time operational parameters, including state of charge (SoC), battery temperature, PV surplus, and degradation indicators. Unlike conventional continuous power control strategies, the proposed approach employs discrete, adaptive charging modes to balance energy utilization and battery health. The EMS is evaluated through a simulation framework using realistic PV generation, load profiles, and thermal conditions. Results demonstrate that the proposed system improves PV self-consumption by approximately 22%, while reducing thermal stress, high-SoC exposure, and unnecessary charge–discharge cycling. These improvements contribute to extending battery lifetime, particularly for second-life battery applications. The findings highlight the effectiveness of integrating machine learning with real-time energy management to achieve efficient, degradation-aware charging in residential and hybrid renewable energy systems</p>

🔗 Provenance — このレコードを発見したソース

🔔 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。