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：
- 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
- 低碳循環之廢棄電池剩餘電能回收裝置研製
- Title
- Design and Implementation of a Residual Energy Recovery Device in Low-Carbon Recycling for Spent Batteries
- 作者
- 徐士賢、林律評、王志祥、張震垠、張育瑋
- Authors
- Shih-Hsien Hsu, Lu-Ping Lin, Jhih-Siang Wang, Chen-Yin Chang, Yu-Wei Chang
- 摘要
- 臺灣每年都會產生大量廢棄電池，根據環境部統計，在111年的廢棄電池回收量達67,686公噸， 112年度增加至70,864公噸，這顯示廢棄電池的回收量相當可觀，而這些電池中仍可能剩餘約50%的 電能(鄭翔文，2020)。因此，若能在分解前萃取並儲存至儲能設備中，將減少廢電池量並提升電池 效益，實現低碳循環。本研究利用神經網路之長短期記憶模型(Long Short-Term Memory, LSTM)、 開路電壓量測(Open Circuit Voltage, OCV)與內阻法(Internal Resistance, IR)進行電池的廢棄判斷與剩 餘電量估測。接著利用放電平衡電路與自適應脈波放電法(Self-Adapt Pulse Discharge, SPD)萃取剩餘 電量，最後使用短路放電法(Short Circuit Discharge, SCD)將所萃取之能量儲存於超級電容中。本研 究透過硬體電路與軟體程式驗證所提方法，適用於一次與二次電池且有效提升萃取效能，並具備人 機介面可顯示電池資訊。原型裝置經實測，LSTM模型進行預測時其平均絕對誤差(Mean Absolute Error, MAE)約為5%，顯示出預測模型的準確性，廢棄電池的實際萃取效率可達約25%，能夠有效 提取剩餘能量，進而提高資源的利用率。本裝置所回收的剩餘電量可供小功率設備運行或為其他二 次電池充電，有助於低碳循環與資源再利用之目的達成。
- 關鍵字
- 廢棄電池，低碳循環，長短期記憶模型，內阻法，自適應脈波放電
- Abatract
- Taiwan generates a significant number of spent batteries annually. The number of recycled used batteries reached 67,686 tons in 2022 and increased to 70,864 in 2023. These discarded batteries may retain approximately 50% of their original energy (Zheng, 2020). Therefore, extracting and storing the remaining energy before disposal can reduce battery waste, enhance efficiency, and contribute to a low-carbon circular economy. This study employs a Long Short-Term Memory (LSTM) neural network model, Open Circuit Voltage (OCV), and Internal Resistance (IR) method to assess battery end-of-life status and estimate remaining energy capacity. A discharge balancing circuit and the Self-Adapt Pulse Discharge (SPD) method are then used to extract the residual energy, which is subsequently stored in a supercapacitor through the Short Circuit Discharge (SCD) technique. The proposed methodology is validated through hardware circuit implementation and software programming, demonstrating its applicability to primary and secondary batteries while effectively improving energy extraction efficiency. The prototype device integrates a humanmachine interface to display battery information. The experimental result shows that the LSTM prediction model can achieve a 5% Mean Absolute Error (MAE), indicating high accuracy, and the practical energy extraction efficiency is around 25%. This device can effectively recover residual energy and enhance resource utilization. The recovered energy can power low-power devices or recharge other secondary batteries, contributing to low-carbon sustainability and resource reutilization.
- Keywords
- Spent Batteries, Low-Carbon Recycling, Long Short-Term Memory (LSTM), Internal Resistance (IR), Self-Adapt Pulse Discharge (SPD).