Abstract:

The reliability and safety of Li-ion batteries (LIBs) are fundamental to the efficacy of battery energy storage systems. Battery capacity degradation and swelling represent critical facets of LIB reliability.  Degradation manifests as a decline in capacity, directly impacting the operational range and performance of batteries. Conversely, swelling denotes the physical expansion of battery cells, a consequence of gas accumulation, heat generation, and alterations within the electrode structure. This phenomenon poses a risk of internal short circuits, potentially precipitating thermal runaway and fire hazards. It becomes essential to elucidate the mechanisms and contributory factors underlying these issues and to devise precise predictive models for these phenomena. This thesis investigates the challenges of degradation and swelling in LIBs, utilizing machine learning (ML) to enhance predictive analytics within battery technology. It seeks to develop advanced predictive models to accurately assess the state-of-health (SoH) and anticipate capacity degradation and swelling issues. Through data analytics and ML techniques, the study constructs models that predict and characterize battery degradation and swelling, supported by a detailed analysis of the factors influencing LIB health and an assessment of various ML methods using real battery data. The initial phase conducts an extensive literature review on ML applications in battery health estimation, evaluating existing methodologies and identifying research gaps. This phase establishes a framework for understanding the factors affecting battery capacity degradation and employs ML algorithms to predict battery degradation. The research advances by developing and validating these algorithms, utilizing empirical data from actual electric vehicle (EV) battery usage, which includes diverse operational conditions and degradation patterns. A novel feature selection technique is employed to identify predictors of battery health, leading to the creation of precise models that forecast degradation. The second phase focuses on the mechanisms of battery swelling through experimental analysis to understand the physical and chemical bases of this phenomenon. This part of the study enhances the empirical validity of the ML predictions, identifying essential factors for battery capacity degradation and swelling. By simulating various operational conditions, the research integrates data such as temperature, State of Charge (SoC), current, voltage, and dimensional parameters into ML algorithms to develop comprehensive models for predicting battery swelling. The thesis evaluates the performance of ML-based models, systematically comparing their predictive reliability to traditional methods and highlighting their capability to predict and mitigate battery failures. These models offer insights into battery health, demonstrating their potential to prolong battery lifespans and enhance battery reliability. The findings provide robust evidence that ML techniques can accurately predict and characterize critical challenges in battery technology. By leveraging advanced ML algorithms, the developed models not only estimate SoH and swelling with high precision but also identify key factors impacting these conditions. This underscores the transformative impact of ML in boosting predictive capabilities and knowledge of LIB health. The research lays a solid foundation for future advances in battery diagnostics and prognostics, offering actionable insights to enhance the reliability of LIBs. This approach contributes to the discourse on battery technology and positions the research at the forefront of efforts to achieve sustainable battery solutions.

Thesis Framework

Final Thesis Defense Committee 

1. Dr. Naveed Arshad (Associate Professor, PhD Supervisor), Department of Computer Science, Lahore University of Management Sciences (LUMS), Pakistan 
2. Dr. Hassan Abbas Khan (Associate Professor, PhD Co-Supervisor, Thesis Committee Member), Department of Electrical Engineering, Lahore University of Management Sciences (LUMS), Pakistan
3. Dr Muhammad Khalid (Associate Professor, PhD Co-Supervisor, Thesis Committee Member), Electrical Engineering Department, King Fahd University of Petroleum and Minerals (KFUPM), Dhahran, Saudi Arabia.
4. Dr. Ijaz Haider Naqvi (Associate Professor, Thesis Committee Member), Department of Electrical Engineering, Lahore University of Management Sciences (LUMS), Pakistan
5. Dr. Tariq Jadoon (Associate Professor, Thesis Committee Member), Department of Electrical Engineering, Provost, Lahore University of Management Sciences (LUMS), Pakistan 
6. Dr. Ali Hussain Kazim (Associate Professor, External Examiner), Department of Mechanical Engineering, University of Engineering and Technology Lahore Pakistan 

List of Publications

1. Rauf, H., Khalid, M., & Arshad, N. (2022). Machine learning in state of health and remaining useful life estimation: Theoretical and technological development in battery degradation modelling. Renewable and Sustainable Energy Reviews (I.F. 16.8), 156, 111903. https://doi.org/10.1016/j.rser.2021.111903
2. Rauf, H., Khalid, M., & Arshad, N. (2023). A novel smart feature selection strategy of lithium-ion battery degradation modelling for electric vehicles based on modern machine learning algorithms. Journal of Energy Storag (I.F. 8.8), e, 68, 107577.  https://doi.org/10.1016/j.est.2023.107577
3. Rauf, H., Khalid, M., Arshad, N., & Pecht, M. (2023, July). Novel Feature Selection Strategy for Cyclic Loss Prediction of Lithium-ion Electric Vehicle Battery. In 2023 IEEE Power & Energy Society General Meeting (PESGM) (pp. 1-6). IEEE. 10.1109/PESGM52003.2023.10253190
4. Rauf, H., Gul, M. S., Khalid, M., & Arshad, N. (2023, August). Smart Feature Selection-Based Machine Learning Framework for Calendar Loss Prediction of Li-Ion Electric Vehicle Battery. In 2023 12th International Conference on Renewable Energy Research and Applications (ICRERA) (pp. 300-303). IEEE. 10.1109/ICRERA59003.2023.10269362
5. Rauf, H., Madipatla, S., Osterman, M. D., Arshad, N., and Pecht, M. “Swelling Mechanisms in Li-ion Batteries,” Under Review in Nature Reviews Material (I.F. 76.8), .
6. Rauf, H., Madipatla, S., Osterman, M. D., Arshad, N., Khalid, M. and Pecht, M. “Experimental Characterization, Investigation and Machine Learning-based Predictive Modeling of Lithium-ion Pouch Cell Battery Swelling” In progress for submission in IEEE Transactions of Industrial Informatics
7. Rauf, H., Khan, H. A., Arshad, N., Pecht, M., & Khalid, M. Reliability and Economic Assessment of Integrated Distributed Hybrid Generation and Battery Storage for Base Transceiver Stations in Intermittent Utility Grids. Under Revision in IEEE Access (I.F 3.5), 

PATENT

1. Rauf, H., Arshad, N., and Khalid, M. (2024). Method and system for predicting battery capacity degradation for electric vehicle. (Accepted; Green Status). U.S. Patent Application No. 550498US (Patent number yet to be allotted)

Other Publications/Projects during PhD:

1. Rauf, H., Gull, M. S., & Arshad, N. (2020). Complementing hydroelectric power with floating solar PV for daytime peak electricity demand. Renewable Energy (I.F 8.8), , 162, 1227-1242.
2. Rauf, H., Gull, M. S., & Arshad, N. (2019). Integrating floating solar PV with hydroelectric power plant: Analysis of Ghazi Barotha reservoir in Pakistan. Energy Procedia, 158, 816-821.
3. Rauf, H., Khan, H. A., & Arshad, N. (2019, November). Optimized power system planning for base transceiver station (BTS) based on minimized power consumption and cost. In 2019 8th International Conference on Renewable Energy Research and Applications (ICRERA) (pp. 773-779). IEEE.