電池使用策略與健康度預測之研究

Abstract

隨著全球環境問題日益嚴峻，鋰離子電池作為電動農機核心動力來源的重要性逐漸提升。本研究探討鋰離子電池的使用策略與健康度預測及老化對電池組性能之影響。透過充電倍率實驗分析LFP、NCM及LTO三種材料電池之充電行為，發現當倍率提升至2 C時，NCM電池溫度上升最高到14.5°C。放電深度實驗顯示，高電量區間循環對容量衰退影響最大，LFP與NCM分別衰退1.75%與2.48%。設定充電SOC上限結果指出，SOC 80%組雖然單次放電容量較低，但整體衰退速率較SOC 100%組減少1.27%，平均充電時間僅為後者之54.69%，適度降低SOC上限有助提升使用效益。研究建立多項式回歸模型預測NCM電池健康度，其中一次、 二次與三次模型之MAE分別為3.95、2.37與0.84。三次多項式雖於訓練資料中表現最佳，但在LOOCV驗證下出現過擬合現象，MAE上升到156.74%顯示泛化能力不足，而二次多項式模型兼具精度與穩健性LOOCV MAE小幅提高至5.98%。老化電池影響實驗顯示，電池組含老化電池時，循環後內阻增加9.31 mΩ，運行溫度提升。研究成果有助鋰離子電池壽命管理與電動農機設計，推動農業電動化及淨零碳排。

As global environmental issues become increasingly severe, the importance of lithium batteries as the core power source for electric agricultural machinery has gained more attention. This study investigates the estimation of lithium battery’s SOH and the effects of aging on battery pack performance. Charge rate experiments were conducted to analyze the charging behavior of three types of batteries: LFP, NCM, and LTO. The results show that when the charge rate increased to 2C, the temperature of the NCM batteries reached up to 14.5°C. Depth of discharge experiments indicate that cycling in the high SOC range has the greatest impact on capacity degradation, with LFP and NCM batteries exhibiting capacity reductions of 1.75% and 2.48%, respectively. In experiments with a set upper limit of charging SOC, despite the SOC 80% group having lower single discharge capacity, its overall degradation rate was 1.27% lower, and its average charging time was only 54.69% compared to the SOC 100% group. These findings suggest that moderately lowering the SOC upper limit can enhance operational efficiency. Polynomial regression models were developed to predict the SOH of NCM batteries, with MAEs of 3.95, 2.37, and 0.84 for first-, second-, and third-order models, respectively. Although the third-order model shows the best performance on the training data, it exhibited overfitting under LOOCV, with its MAE rising to 156.74, indicating poor generalizability, whereas the second-order model balances accuracy and robustness, with its MAE under LOOCV increasing only marginally to 5.98%. Experiments examining the effects of aged cells reveal that including aged batteries in the pack leads to a sharp increase in internal resistance by 9.31 mΩ after cycling, accompanied by elevated operating temperatures. The findings of this study provide valuable insights for lithium battery life management and the design of electric agricultural machinery, thereby advancing agricultural electrification and net-zero carbon emissions.