Anomaly diagnosis and health monitoring of lithium-ion battery packs

Restricted (Penn State Only)
- Author:
- Bhaskar, Kiran
- Graduate Program:
- Mechanical Engineering
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- June 12, 2024
- Committee Members:
- Robert Kunz, Professor in Charge/Director of Graduate Studies
Daning Huang, Outside Unit & Field Member
Herschel Pangborn, Major Field Member
Satadru Dey, Major Field Member
Chris Rahn, Chair & Dissertation Advisor - Keywords:
- Fault detection
Battery pack
Battery safety
Parallel packs
Cell sorting
Fault diagnosis
State of charge
State of health
Temperature distribution - Abstract:
- In the last two decades, the growing power demands and transportation electrification popularized lithium-ion batteries for energy storage applications, such as the power grid, electric vehicles, and electric locomotives, due to their long cycle life, extended calendar life, low self-discharge rate, and high energy and power density. However, safety has been an increasing concern with the rising growth of lithium-ion batteries. The anomalies in lithium-ion batteries are often caused by electrical abuse, thermal abuse, mechanical abuse, manufacturing defects, and internal degradation due to aging. Thus, early detection and mitigation of anomalies are critical to avoid fault propagation, ensuring safe and reliable operation of lithium-ion battery packs, improving performance, and ensuring the safety of lithium-ion battery packs. Furthermore, these battery packs consist of hundreds of battery cells connected in series and parallel to meet the voltage and capacity requirements. The heterogeneity in battery packs limits the battery performance and can accelerate battery degradation. Hence, it is essential to understand the battery pack dynamics and quantify the pack heterogeneity to improve the battery performance and longevity of these packs. By leveraging cell-to-cell similarity and heterogeneity in battery packs, this dissertation focuses on developing anomaly diagnosis and health monitoring strategies, using model-based and data-driven techniques, to improve the safety, performance, and longevity of lithium-ion battery packs. First, offline and real-time State of Health estimation approaches, coupled with a State of Charge observer, are developed, which are further extended to identify and quantify micro short circuits in battery packs. Additionally, a data-driven anomaly detection approach is presented to accurately detect and classify voltage and temperature anomalies in battery packs. Second, this dissertation proposes signal reconstruction strategies to deal with sensor-related anomalies. Using a data-driven approach, anomalous voltage and temperature sensor signals are reconstructed using the rest of the nominal voltage and temperature measurements. Third, to identify the local hot spots that could potentially cause safety concerns, the 2D temperature distribution of the pouch cells is estimated in a sparse temperature sensing scenario without prior knowledge of physics-based models and model parameters. An optimal temperature sensor placement strategy to extract maximum thermal information from sparse sensing scenarios is also proposed. Fourth, this dissertation focuses on understanding and predicting the effect of parameter heterogeneity in parallel-connected cell groups, leading to capacity and power loss. Therefore, this dissertation develops multiple anomaly diagnosis and health monitoring strategies incorporating the effect of cell heterogeneities in lithium-ion battery packs, working towards safe, reliable, and durable energy storage systems.