臺灣能源期刊發行

  • 創刊日期：

    102年11月30日

  • 發行所：

    經濟部能源署

  • 發行人：

    李君禮

  • 地址：

    台北市復興北路2號13樓

  • 電話：

    02-2772-1370

  • 執行單位：

    財團法人工業技術研究院

  • 地址：

    新竹縣竹東鎮中興路四段195號26館

  • 服務專線：

    03-5916006

  • 服務信箱：

    JOET@itri.org.tw

  • 總編輯：

    劉志文胡均立

  • 顧問：

    王漢英童遷祥

  • 執行主編：

    劉子衙陳志臣

  • 編輯委員：

    方良吉王錫福朱家齊李堅明李叢禎林師模馬鴻文陳希立廖芳玲廖肇寧劉文獻蕭志同顧洋(依筆畫順序排列)

:::

臺灣能源期刊論文全文

臺灣能源期刊第12卷第1期內容

出刊日期：March, 2025

題目
數據驅動主從式充放電策略之低碳車對網交易研製
Title
Development of a Data-Driven Master-Slave Charging and Discharging Strategy for Low-Carbon Grid-to-Vehicle (G2V) Trading Applications
作者
黃俊瑋、林星妤、徐士賢、李秉昕
Authors
Chun-Wei Huang, Hsing-Yu Lin, Shih-Hsien Hsu, Ping-Hsin Lee
摘要
隨著低碳運輸轉型，運具電力化成為了不可忽略的趨勢。在減碳方面，主要有電動車(EV)、插 電式混合動力車(PHEV)、氫燃料電池車(FCV)和油電混合車(Hybrid V)等四種低碳運輸工具，而這 四種減碳效益差異較大，主要取決於其動力系統、能源來源、能源效率和使用環境等因素，其中以 純電動車與氫燃料電動車的減碳效益最大，但因為氫燃料之成本過於高昂，因此目前純電動車為世 界之浪潮。但純電動車極度依賴清潔電力，因此，電池管理系統(Battery Management System, BMS) 於電動車充放電管理上扮演起至關重要之角色，負責監控碳排放流、負載潮流與電力資訊的關鍵訊 息。因此，本研究旨在建構電動車的充放電策略，調度電池充電狀態監控模式，並通過控制主從式 電池組實現充放電狀態排程與市電併網調度。同時還設計低碳車對網交易模型，持續監控電力碳足 跡、碳費計算與剩餘容量估計，俾利在充放電策略中執行最佳決策。本研究開發電動車充放電策略 指引，並通過研究結果篩選ANFIS、GRU、LSTM、GRU-LSTM、LSTM-GRU等5種剩餘容量估測 模型分析比較，以優化其估測性能。研究中建立不同環境溫度與充放電速率情境，並於同一電池容 量(200 Ah)下進行驗證。研究成果顯示主電池平均消耗為62.75 Ah、副電池平均消耗為71 Ah時，單 電池的平均消耗為173 Ah。實驗數據驗證了電池管理系統的優化效果，其中LSTM-GRU模型的估測 性能評價指標(MAE)為3.56%，展現了卓越的追蹤能力。此外，在碳排放量的分析中，透過計算碳 排放量與碳費，展示碳排之控制成效。與單電池相比，主從式充放電策略的主副電池碳排放量平均 少了0.06 kgCO2e。最後，實驗結果表明該策略在實際硬體環境中的可行性，並有望在實際應用中為 里程焦慮提供有效的解決方案。
關鍵字
主從式充放電策略排程，電池電量狀態，碳排監測
Abatract
With the transformation of low-carbon transportation, the electrification of transportation vehicles has become a trend that cannot be ignored. In terms of carbon reduction, there are mainly four low-carbon transportation vehicles such as electric vehicles (EV), plug-in hybrid vehicles (PHEV), hydrogen fuel cell vehicles (FCV) and hybrid vehicles (Hybrid V), and these four carbon reduction benefits are quite different, mainly depending on their power system, energy source, energy efficiency and use environment and other factors, among which pure electric vehicles and hydrogen fuel electric vehicles have the greatest carbon reduction benefits, but because the cost of hydrogen fuel is too high, pure electric vehicles are currently the wave of the world. However, BEVs are extremely dependent on clean electricity, so the battery management system (BMS) plays a crucial role in EV charging and discharging management. Within this context, the Battery Management System (BMS) plays a pivotal role in managing the charging and discharging of electric vehicles, specifically by monitoring key information such as carbon emissions flow, load flow, and power data. Consequently, this study aims to develop charging and discharging strategies for electric vehicles, facilitating the master-slave battery packs for charge/discharge scheduling and grid connection. Additionally, a low-carbon grid-to-vehicle (G2V) trading model was designed, with continuous monitoring of the energy-related carbon footprints, carbon cost calculations, and state-of-charge estimation to enable optimal decision-making in the charging and discharging strategy. This study develops a state of charge (SOC) estimation guideline for low-carbon electric vehicle charging and discharging strategies. Based on the research results, five state-of-charge estimation models, ANFIS, GRU, LSTM, GRU-LSTM, and LSTM-GRU, were be selected for analysis and comparison to optimize their estimation performance. Different ambient temperature and (dis)charge rate scenarios were established to optimize the Battery Management System (BMS). Under verification with the same battery capacity (200 Ah), the results showed that the average consumption was 62.75 Ah for the primary battery and 71 Ah for the secondary battery, compared to 173 Ah for the single battery. Experimental data validated the optimization of the Battery Management System, with the LSTM-GRU model demonstrating superior estimation performance, achieving a Mean Absolute Error (MAE) of 3.56%, indicative of excellent tracking capability. Furthermore, the analysis of carbon emissions, including the calculation of carbon emissions and carbon costs, demonstrated effective carbon control. Compared to single-battery operation, the masterslave charging/discharging strategy reduced the average carbon emissions by 0.06 kgCO2e. Finally, the experimental results confirm the feasibility of this strategy in actual hardware environments, offering a promising solution for mitigating range anxiety in practical applications.
Keywords
Master-Slave Charging and Discharging Strategy, Battery State of Charge, Carbon Emission Monitoring.
全文下載/Download