Lithium-Ion Battery Life Prediction with Early Cyclic Data Using Machine Learning

Abstract

To accomplish the preliminary estimation of profit-oriented battery longevity, this research utilized machine learning techniques. We calculated the number of correct predictions to the battery dataset of various machine learning techniques. The classification tree (CT) approach showed the maximum and best precision of 98.2 percent among multiple algorithms to forecast whether the battery will sustain over 90 percent primary power after 660 cycles. Utilizing the preliminary two data periods, CT suggests that the primary function for calculating the durability of batteries is the difference in discharge power. Given the initial 200 cycles, the peak weight factor switches to the internal resistance for calculating battery’sdurability.

Introduction

I. Introduction

Lithium-ion (Li-ion) batteries are central to technological innovation, but predicting their lifespan remains complex. Traditional empirical and physical models are often inadequate due to:

• Non-linear degradation processes,

• Inconsistency in performance,

• And the dynamic internal structure of the batteries.

Modern approaches now use machine learning (ML) to predict battery life by analyzing early cycle data and electrochemical behavior.

II. Related Work

Researchers have explored:

• Empirical & diffusion models

• Semi-empirical methods

• Data-driven ML techniques

Notable studies include:

• Shen et al. (2020): Integrated ML with empirical frameworks using 83 battery samples.

• Qin et al. (2019): Used Modelica for battery simulation with <2% error.

• Saidani et al. (2017): Compared various Li-ion types using observational data.

• Zhang et al. (2019): Used recovery phenomena for life prediction.

• Ramadesigan et al. (2012): Focused on systems-level modeling of battery behavior.

III. Proposed Work

ML-Based Prediction Framework

The approach consists of three phases:

1. Data Collection & Feature Selection

2. Model Development

3. Evaluation

• Data collected from High Power Nanophosphate Li-ion cells using Arbin testing equipment.

• Batteries classified into:

  • Strong longevity (>660 cycles)

  • Weak longevity (<660 cycles or <90% capacity)

Feature Engineering

• Used data from early cycles (1st, 2nd, up to 200th).

• Key features:

  • Charge/Discharge capacity (C-1, D-1)

  • Internal resistance (IR-1)

  • Temperature (T-1)

  • Differences between early cycles (e.g., C2–1, D2–1)

• Created feature sets such as “2-1”, “5-1”, … “200-1”.

ML Models Used

• Classification Tree (CT) (based on CART)

• Random Forest

• AdaBoost

• Naive Bayes

• SVM, k-NN, and other standard classifiers

IV. Results & Analysis

1. Correlation Analysis

• Most single features showed weak correlation (Pearson’s r < 0.3) with battery life.

• Inner resistance and capacity difference between early cycles had stronger predictive power.

2. ML Model Performance

• CT with “2-1” feature achieved 95% accuracy

• Random Forest: 92%

• AdaBoost: 90.5%

• Naive Bayes: 78.6%

• CT remained effective across various cycle sets (e.g., 5th, 10th, 100th).

3. Feature Importance Insights

• In early cycle analysis, discharge capacity (D2-1) was the most informative.

• For longer cycles, internal resistance (IR100-1) became the dominant predictor.

• Shows that as batteries age, internal resistance is a critical degradation indicator.

V. Conclusions

• Early cycle data (as little as the first 2 cycles) can effectively predict battery lifespan using ML.

• Classification Trees (CT) offer high accuracy and interpretability.

• Internal resistance is a key predictor, especially in long-term performance.

• Scaling this method requires larger datasets and further validation across battery chemistries.

Key Takeaways

• Machine learning outperforms traditional methods for Li-ion lifespan estimation.

• Internal resistance and early cycle behavior are vital predictive features.

• CT models offer both high accuracy and clear interpretability, making them suitable for practical battery management systems.

Conclusion

Finally, the CT algorithm was utilized to predict the Li-Ion battery longevity. In order to obtain the best possible outcome it was observed that the predictive performance of the different ML methods was correlated with the battery dataset. Identified accuracy will reach 98.2 percent only with statistics collected from the initial two cycles of Li-Ion batteries. The most critical consideration for Li-Ion cell life has been calculated by interpreting the CT as the gap in the discharged power around two preceding cycles (D2-1 and D-1). The tremendous promise of ML and data processing in battery based technology is simulated in this paper.

References

[1] Shen, S., Nemani, V., Liu, J., Hu, C., & Wang, Z. (2020, June). A Hybrid Machine Learning Model for Battery Cycle Life Prediction with Early Cycle Data. In 2020 IEEE Transportation Electrification Conference & Expo (ITEC) (pp. 181-184). IEEE. [2] Qin, D., Li, J., Wang, T., & Zhang, D. (2019). Modeling and Simulating a Battery for an Electric Vehicle Based on Modelica. Automotive Innovation, 2(3), 169-177. [3] Saidani, F., Hutter, F. X., Scurtu, R. G., Braunwarth, W., & Burghartz, J. N. (2017). Lithium-ion battery models: a comparative study and a model-based powerline communication. Advances in Radio Science: ARS, 15, 83. [4] Zhang, Z., Shen, D., Peng, Z., Guan, Y., Yuan, H., & Wu, L. (2019). Lithium-ion batteries Remaining Useful Life Prediction Method Considering Recovery Phenomenon. Int. J. Electrochem. Sci, 14, 7149-7165. [5] Ramadesigan, V., Northrop, P. W., De, S., Santhanagopalan, S., Braatz, R. D., & Subramanian, V. R. (2012). Modeling and simulation of lithium-ion batteries from a systems engineering perspective. Journal of the electrochemical society, 159(3), R31. [6] Zhang, Z., Peng, Z., Guan, Y., & Wu, L. (2020). A Nonlinear Prediction Method of Lithium-Ion Battery Remaining Useful Life Considering Recovery Phenomenon. Int. J. Electrochem. Sci, 15, 8674-8693. [7] Xu, X., Yu, C., Tang, S., Sun, X., Si, X., & Wu, L. (2019). Remaining useful life prediction of lithium-ion batteries based on Wiener processes with considering the relaxation effect. Energies, 12(9), 1685. [8] Cong, X., Zhang, C., Jiang, J., Zhang, W., Jiang, Y., & Jia, X. (2020). An Improved Unscented Particle Filter Method for Remaining Useful Life Prognostic of Lithium-ion Batteries With Li (NiMnCo) O 2 Cathode With Capacity Diving. IEEE Access, 8, 58717-58729. [9] Zhao, L., Wang, Y., & Cheng, J. (2019). A hybrid method for remaining useful life estimation of lithium-ion battery with regeneration phenomena. Applied Sciences, 9(9), 1890. [10] Cong, X., Zhang, C., Jiang, J., Zhang, W., & Jiang, Y. (2020). A Hybrid Method for the Prediction of the Remaining Useful Life of Lithium-ion Batteries with Accelerated Capacity Degradation. IEEE Transactions on Vehicular Technology, 69(11), 12775-12785. [11] Xie, G., Peng, X., Li, X., Hei, X., & Hu, S. (2020). Remaining useful life prediction of lithium?ion battery based on an improved particle filter algorithm. The Canadian Journal of Chemical Engineering, 98(6), 1365-1376. [12] Chen, L., An, J., Wang, H., Zhang, M., & Pan, H. (2020). Remaining useful life prediction for lithium-ion battery by combining an improved particle filter with sliding-window gray model. Energy Reports, 6, 2086-2093. [13] Wu, Z., Wang, Z., Qian, C., Sun, B., Ren, Y., Feng, Q., & Yang, D. (2019, March). Online prognostication of remaining useful life for random discharge lithium-ion batteries using a gamma process model. In 2019 20th International Conference on Thermal, Mechanical and Multi-Physics Simulation and Experiments in Microelectronics and Microsystems (EuroSimE) (pp. 1-6). IEEE. [14] Chen, L., Wang, H., Chen, J., An, J., Ji, B., Lyu, Z., ... & Pan, H. (2020). A novel remaining useful life prediction framework for lithium?ion battery using grey model and particle filtering. International Journal of Energy Research. [15] Deng, L., Shen, W., Wang, H., & Wang, S. (2020). A rest-time-based prognostic model for remaining useful life prediction of lithium-ion battery. Neural Computing and Applications, 1-12. [16] Zhang, S., Guo, X., Dou, X., & Zhang, X. (2020). A rapid online calculation method for state of health of lithium-ion battery based on coulomb counting method and differential voltage analysis. Journal of Power Sources, 479, 228740. [17] Zhu, Shan, Naiqin Zhao, and Junwei Sha. \"Predicting battery life with early cyclic data by machine learning.\" Energy Storage 1.6 (2019): e98. [18] Fei, Zicheng, et al. \"Early prediction of battery lifetime via a machine learning based framework.\" Energy 225 (2021): 120205. [19] Yang, Yixin. \"A machine-learning prediction method of lithium-ion battery life based on charge process for different applications.\" Applied Energy 292 (2021): 116897. [20] Celik, Belen, et al. \"Prediction of battery cycle life using early-cycle data, machine learning and data management.\" Batteries 8.12 (2022): 266.

Copyright

Copyright © 2025 V T Ram Pavan Kumar M, Korikana Gayatri, K V R Divya Madhuri, Ch. Sindhu, S Lavanya, Devara Naga Anusha. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.