鋰電池容量與衰退的傳統預測方法，如過度依賴充放電歷史數據，存在模型透明度不足與預測時效性不佳等問題。本研究提出一種方法，能於早期階段篩選出影響鋰電池容量與衰退的關鍵因子。研究方法採用一種可生成模式樹的演算法，其具有決策節點進行資料切分，葉節點則為線性迴歸模型以進行預測。所蒐集的特徵資料來自製造與檢驗階段，包含正負極極片的延伸量、厚度、塗佈量、含水率、導電度與放置時間。預測的三個目標變數為：放電容量、充電容量與庫倫效率。資料前處理步驟包括特徵篩選、正規化與共線性分析，模型評估則採用十折交叉驗證。實驗結果顯示，針對放電容量與充電容量所建構的模式樹模型，其相關係數分別為0.5145與0.5307，相對絕對誤差約為85%，顯示具有中等程度的預測能力。兩棵模式樹所識別出的關鍵影響變數為：正極放置時間、負極烘烤後放置時間、正極塗佈量，以及電解液含水率。根據這些發現，提出可行建議，包括縮短材料放置時間、優化乾燥條件與強化原料控管。而庫倫效率模型的相關係數低於0.1，顯示其關鍵因子仍需進一步探討與篩選。

Conventional methods for estimating lithium-ion battery capacity and degradation, such as over-reliance on charge/discharge history data, are lack of model transparency and insufficient timeliness of predictions. This study proposes an approach that can filter critical factors for lithium-ion battery capacity and degradation in an early stage. The methodology adopts an algorithm to grow model trees that have decision nodes for splitting and leaf nodes with linear regression models for prediction. The features collected from manufacturing and inspection stages include elongation, thickness, coating weight, moisture content, conductivity, and dwell time of cathode and anode electrodes. Three target variables for prediction are discharge capacity, charge capacity, and coulombic efficiency. The preprocessing tools applied on the collected data includes feature selection, normalization, and collinearity analysis, and 10-fold cross-validation is the method for performance evaluation. The experimental results show that the model trees for discharge and charge capacity achieve correlation coefficients 0.5145 and 0.5307, respectively. The relative absolute errors in both cases are around 85%, indicating moderate predictive capability. Key influencing variables identified from the two model trees are cathode dwell time, post-oven anode dwell time, cathode coating weight, and electrolyte moisture content. Based on these findings, actionable suggestions are proposed, including reducing material dwell times, optimizing drying conditions, and enhancing raw material control. The correlation coefficient of the model tree for coulombic efficiency is less than 0.1 so that further study is necessary for filtering its key factors.

Biao, Z. Z. (2023). Design and realization of data mining simulation and methodological models. Journal of King Saud University-Science, 35(10), 102964.
Brunoe, T. D., Andersen, R., Nielsen, K., & Andersen, A. L. (2024). Data mining for characterizing manufacturing capabilities. Procedia CIRP, 130, 220-225.
Chen, J. (2021). Application of data mining technology in software engineering. In Application of Intelligent Systems in Multi-modal Information Analytics: 2021 International Conference on Multi-modal Information Analytics (MMIA 2021), Volume 2 (pp. 346-351). Springer International Publishing.
Duque, J., Godinho, A., Moreira, J., & Vasconcelos, J. (2024). Data Science with Data Mining and Machine Learning A design science research approach. Procedia Computer Science, 237, 245-252.
Guo, Z., Li, Y., Yan, Z., & Chow, M. Y. (2024). A Neural-Network-Embedded Equivalent Circuit Model for Lithium-ion Battery State Estimation. arXiv preprint arXiv:2407.20262.
Han, H. L., Ma, H. Y., & Yang, Y. (2019). Study on the test data fault mining technology based on decision tree. Procedia Computer Science, 154, 232-237.
Lei-da Chen, T. S., & Frolick, M. N. (2000). Data mining methods, applications, and tools. Information systems management, 17(1), 67-68.
Lu, F., Jin, Z., & Zhang, X. (2023). An Analysis of Data Mining Techniques in Software Engineering Database Design. Procedia Computer Science, 228, 47-56.
Lei, L., Wu, B., Fang, X., Chen, L., Wu, H., & Liu, W. (2023). A dynamic anomaly detection method of building energy consumption based on data mining technology. Energy, 263, 125575.
Lv, X., Cheng, Y., Ma, S., & Jiang, H. (2024). State of health estimation method based on real data of electric vehicles using federated learning. International Journal of Electrochemical Science, 19(8), 100591.
Long, X., & Chen, W. (2024). Construction framework of smart tourism big data mining model driven by blockchain technology. Heliyon, 10(14).
Miao, D., Lv, Y., Yu, K., Liu, L., & Jiang, J. (2023). Research on coal mine hidden danger analysis and risk early warning technology based on data mining in China. Process Safety and Environmental Protection, 171, 1-17.
Paccha-Herrera, E., Jaramillo-Montoya, F., Calderón-Muñoz, W. R., Tapia-Peralta, D., Solórzano-Castillo, B., Gómez-Peña, J., & Paccha-Herrera, J. (2024). A particle filter-based approach for real-time temperature estimation in a lithium-ion battery module during the cooling-down process. Journal of Energy Storage, 94, 112413.
Szramowiat-Sala, K., Penkala, R., Horák, J., Krpec, K., Hopan, F., Ryšavý, J., ... & Górecki, J. (2024). AI-based data mining approach to control the environmental impact of conventional energy technologies. Journal of Cleaner Production, 472, 143473.
Wei, Y., Yan, Y., Zhang, C., Meng, K., & Xu, C. (2023). State estimation of lithium-ion batteries based on the initial rise time feature of ultrasonic signals. Journal of Power Sources, 581, 233497.
Zhang, X., Xiang, H., Xiong, X., Wang, Y., & Chen, Z. (2024). Benchmarking core temperature forecasting for lithium-ion battery using typical recurrent neural networks. Applied Thermal Engineering, 248, 123257.
王昱鈞（2023）。基於殘差卷積神經網絡之鋰離子電池容量狀態預估之研究，國立清華大學工程與系統科學系，碩士論文。
白紘瑜（2023）。基於時域輔助解耦遞迴最小平方法之線上鋰離子電池等效電路模型參數識別和電量狀態估測之研究，國立臺灣科技大學電機工程系，博士論文。
江俊佑、劉雯（2024）。基於Bayes-Deformer模型預測鋰電池的健康狀態。智慧科技與應用統計學報，21(2)，39-62。
吳宗諺（2023）。基於等效電路參數分析鋰電池健康狀態。電腦與通訊，(194)，55-60。
吳涌睿（2023）。結合特斯拉閥特性之循環式熱管改良鋰離子電池散熱系統，國立勤益科技大學機械工程系，碩士論文。
林君宜（2015）。運用M5'模式樹分析放款發生逾期因子。國立成功大學EMBA，碩士論文。
徐永豐、曾令明（2013）。階層性多項歷程樹狀模式在記憶缺陷評估之應用：以台灣臨床資料分析為例。中華心理學刊，55(1)，57-73。
徐晟峻、翁志祁（2008）。鋰電池放電行為高斯－多項式的三角函數溫度模型。華岡工程學報，(22)，193-197。
康鈞誌（2009）。運用模式樹建構半導體測試時間之診斷模式。國立成功大學工業與資訊管理學系碩士在職專班，碩士論文。
張芳瑜（2023）。開發應用於高安全性鋰離子電池之新穎具精準活化溫度之高分子添加助劑PDA-BMI-Li/Trixene BlockedIsocyanates，國立臺灣科技大學化學工程系，碩士論文。
張舜長、徐永佾（2013）。類神經網路應用於電池殘電量估測之研究。車輛工程學刊，(10)，59-73。
張憲國、李政達、陳志弘、劉勁成、陳蔚瑋（2018）。臺中港類神經與M5颱風波浪推算模式之比較。港灣季刊，(109)，32-42。
郭宏明（2006）。運用頻繁模式樹改善小型品項儲位配置與揀貨效率之應用。東海大學資訊工程與科學系碩士在職專班，碩士論文。
程一鵬（2008）。使用模式樹建構SMT錫膏印刷製程品質管制模式之研究。國立成功大學工業與資訊管理學系專班，碩士論文。
黃建翔（2015）。資料探勘在教育領域之發展與應用。臺灣教育評論月刊，4(8)，78-84。
黃康哲（2015）。鋰電池充放電外殼應變量測。三聯技術，(97)，9-13。
黃嘉賢、張舜長、周明正、劉鈞宇（2009）。鋰電池殘電量偵測顯示器研製之應用於複合電動車輛。車輛工程學刊，(6)，103-121。
黃誌成（2017）。應用模式樹與偏最小平方法預測氧化銦錫薄膜電阻值之研究。國立成功大學工業與資訊管理學系碩士在職專班，碩士論文。
楊健杰、翁志祁（2012）。鋰離子電池放電的二次高斯模型。華岡工程學報，(29)，159-164。
廖啓荏（2017）。使用頻繁模式樹預測用電行為的智能插座。國立雲林科技大學電機工程系，碩士論文。
謝孟書（2014）。應用模式樹比較不動產之實價與銀行鑑價之研究。國立成功大學工業與資訊管理學系碩士在職專班，碩士論文。