Bridging AI and Battery Innovation: A Spotlight on the Recent Publication from ENERGETIC Researchers
The ENERGETIC project is proud to celebrate a key milestone in its mission to revolutionize battery management systems (BMS) and contribute to the development of next-generation battery technologies. A recent research article, AI-enabled thermal monitoring of commercial (PHEV) Li-ion pouch cells with Feature-Adapted Unsupervised Anomaly Detection published in the Journal of Power Sources, exemplifies the innovative advancements being pursued within the scope of ENERGETIC.
This study, authored by Abdelrahman Shabayek, Arunkumar Rathinam, Matthieu Ruthven, Djamila Aouada, and Tazdin Amietszajew, addresses a critical challenge in the battery sector: ensuring reliable thermal monitoring to optimize performance and safety. As part of the broader ENERGETIC vision, this work aligns seamlessly with the project’s focus on integrating cutting-edge artificial intelligence (AI) techniques with enhanced sensing technologies to improve battery utilization in both first-life (transport) and second-life (stationary) applications.
Why Thermal Monitoring Matters for Battery Management
Lithium-ion batteries, particularly in plug-in hybrid electric vehicles (PHEVs), are subjected to high-stress environments that can lead to degradation, reduced performance, and safety risks. Effective thermal monitoring plays a pivotal role in enhancing thermal management strategies and minimizing these risks. This is especially significant in the context of ENERGETIC, where one of the project’s objectives is to create a comprehensive digital twin of the battery system.
The Innovation: Feature-Adapted Unsupervised Anomaly Detection (FAUAD)
At the heart of this research is the development of a novel AI model, Feature-Adapted Unsupervised Anomaly Detection (FAUAD). This methodology characterizes the normal behavior of thermal data obtained from Li-ion pouch cells and detects anomalies with remarkable precision. The model leverages deep learning to process thermal data captured via infrared imaging, a technology that ensures detailed temperature profiling across the battery’s surface area.
Key highlights of the FAUAD model include:
- Versatility: The model is agnostic to battery cell chemistry, making it adaptable to a wide range of applications.
- High Performance: FAUAD achieved exceptional anomaly detection results:
- Area Under the ROC Curve (AUROC) score of 0.971 on simulated data.
- AUROC score of 0.990 on contaminated real-world data.
- A perfect AUROC score of 1.0 on clean real-world data.
Implications for ENERGETIC and Beyond
This publication underscores the ENERGETIC project’s commitment to harnessing advanced AI technologies to address critical challenges in battery management. The insights gained from this study are important for :
- Monitoring and Prognosis: Enhancing the ability to predict the remaining useful life of Li-ion batteries with precision.
- Diagnostics and Explainability: Using explainable AI models to delve deeper into the root causes of battery degradation.
- Real-World Application: Supporting safer, more reliable, and powerful battery operations in both transport and stationary use cases.
A Step Closer to ENERGETIC’s Vision
The success of this research marks a significant achievement for the ENERGETIC project. It not only advances the scientific understanding of thermal behavior in Li-ion batteries but also demonstrates the practical applicability of AI-driven solutions in real-world scenarios. As ENERGETIC continues to push the boundaries of innovation, this milestone serves as both a testament to the team’s expertise and a glimpse into the transformative potential of next-generation battery technologies.
Stay tuned for more updates as we continue to drive progress towards a safer, more sustainable future.