Cloud and digital-twin enhanced thermal safety framework for e-mobility battery management systems

Lithium-ion batteries (LIBs) are the preferred energy storage for electric vehicles (EVs) due to their high energy density and efficiency. However, frequent EV fires expose critical shortcomings in current battery management systems (BMS) and thermal controls. LIBs exhibit nonlinear characteristics sensitive to operating and environmental conditions, but existing BMS technologies rely solely on surface temperature measurements, neglecting core temperatures, which can be up to 10°C higher during fast charging or high-load conditions. This limitation contributes to overheating, accelerated degradation, and thermal runaway. This thesis develops advanced core temperature estimation techniques, including hybrid equivalent circuit models (ECM), long short-term memory (LSTM) networks, bidirectional LSTM (Bi-LSTM), and Kolmogorov-Arnold network–LSTM fusion networks. These methods achieve real-time core temperature estimation with errors as low as 0.16°C across a wide range of ambient temperatures and charging/discharging C-rates. A digital twin (DT)-based core temperature prediction model forecasts temperatures three minutes ahead with a prediction error under 0.4°C. Validation includes LIB cells of varying form factors, such as 18650 and 21700, and chemistries like nickel cobalt aluminum oxide (NCA) and nickel manganese cobalt oxide (NMC). To enhance scalability and functionality, the research integrates cloud computing and DT technologies for real-time monitoring, predictive control, and efficient thermal management. Internal temperature-informed closed-loop control demonstrated effective overheating prevention, improving response time by two minutes. Unlike previous studies, this research validates the framework using a 14-cell LIB module, rather than single cells. A key innovation is the DT-based BMS, which predicts thermal behavior up to three minutes in advance, aligning with IEC 62933-2-1:2017 standards for proactive risk mitigation. Cloud-based monitoring and data storage support predictive maintenance and second-life applications for retired batteries. The integration of machine learning, Internet of Things (IoT), and DT technologies ensures adaptability to dynamic conditions. Computational cost and latency were analyzed, with latencies ranging from 5 ms locally to 85 ms for cloud-based systems. By addressing critical gaps in BMS capabilities and introducing predictive thermal management strategies, this research advances EV battery safety and management.

Name thesis:degree_name

Doctor of Philosophy (PhD)

Discipline thesis:degree_discipline

Electrical and Computer Engineering

Grantor

University of Ontario Institute of Technology

Year dc:date.issued

2025

Author dc:creator

• Samanta, Akash

Advisor dc:contributor.advisor

• Williamson, Sheldon

Language dc:language.iso

en

Handle dc:identifier.uri

https://hdl.handle.net/10155/1937

OAI identifier oai:identifier

oai:ontariotechu.scholaris.ca:10155/1937