Please click the link below to join the webinar:

https://kth-se.zoom.us/j/66394762814

Detailed agenda:

14:00-14:30: Data-driven battery modelling using LPV models (Tijs Donkers)

Abstract:
In this talk, I will present our latest results on data-driven battery modelling using linear parameter-varying models. In general, any data-driven modelling activity can be divided into the following steps: (i) selection of proper model structure, (ii) acquiring an informative identification dataset, and (iii) choosing an appropriate parameter estimation procedure. I will present the considerations that we made on each of these steps and I will show and reflect on the experimental results that we have obtained for NMC and LFP cells.

Bio:
Tijs Donkers is an Associate Professor in the Control Systems group of the Dept of Electrical Engineering of Eindhoven University of Technology, Netherlands. His current research focusses on developing control theory specifically for automotive applications, with an emphasis on the computational aspects of the control algorithms. It covers several aspects of modelling and control of batteries (such as optimal charging, cell balancing and state estimation) and distributed optimization and (model predictive) control for (complete) vehicle energy management and ecodriving.

14:30-15:00: Model-Based Evaluation of Cell Heterogeneities and Module Configurations in Parallel-Connected Battery Modules: Enhancing Performance and Aging Homogeneity (Davide Raimondo)

Abstract:
Higher energy capacity in lithium-ion battery systems is achieved by connecting tens to hundreds of cells in parallel. However, managing large battery packs becomes challenging due to cell-to-cell variations, leading to imbalances in state of charge and thermal gradients. These variations can accelerate degradation processes and pose safety concerns. In this work, a sensitivity analysis evaluates CtC heterogeneities’ impact on module performance, considering electrochemical parameters and module setups. Results show manufacturing-dependent parameters, especially electrode thickness, significantly affect module capacity, energy, and current distribution. Module design configuration also influences temperature and thermal gradient, impacting long-term degradation. To mitigate thermal gradients, a simple cell arrangement strategy leveraging manufacturing-induced CtC variations is proposed, reducing aging gradient by up to 300%.

Bio:
Davide M. Raimondo earned his Ph.D. in Electronics, Computer Science, and Electrical Engineering from the University of Pavia, Italy, in 2009. Following this, he embarked on a Postdoctoral Fellowship at the Automatic Control Laboratory, ETH Zürich, Switzerland, from 2009 to 2010. He subsequently served as an Assistant Professor at the University of Pavia from 2010 to 2015, and later as an Associate Professor from 2015 to 2021. In December 2021, he was appointed as a Full Professor at the same institution, a position he held until September 2023. Since October 2023, Dr. Raimondo has assumed the role of Full Professor at the University of Trieste. Over the course of his career, Dr. Raimondo has also held visiting positions at several institutions including MIT, USA; the University of Seville, Spain; TU Wien, Austria; and the University of Konstanz, Germany. Dr. Raimondo is the author or co-author of more than 100 papers published in refereed journals, books, and conference proceedings. His primary research interests encompass a wide array of topics, including battery management systems, set-based estimation, fault diagnosis, fault-tolerant control, model predictive control, and optimization. Dr. Raimondo has received prestigious awards including the Automatica Paper Prize and serves as a Subject Editor for Automatica and IEEE Transactions on Control Systems Technology. He has been an IEEE senior member since 2022.

15:00-15:30 Break

15:30-16:00: Battery modelling, health diagnosis and lifetime prognosis: The ML edge (Changfu Zou)

Abstract:
In this presentation, we will discuss how machine learning (ML) can be harnessed to advance battery modelling, health diagnosis and lifetime prognosis. The first part will be model-integrated neural networks (MINN) for battery modelling. While existing models for battery management often trade accuracy or physical insights for computational efficiency, we propose a new physics-based learning architecture, termed MINN, to develop battery models that are physically insightful, numerically accurate, and computationally tractable. The second part will be ML-based diagnosis and prognosis of battery ageing under arbitrary vehicle usage conditions. We introduce a joint health-lifetime estimation framework that incorporates offline-developed global models, an online adaptation algorithm, and Kalman filter-based model fusion. This framework is designed to work with both time-series and histogram data, and offer a more accurate and practical approach to battery ageing estimation.

Bio:
Changfu Zou is an Associate Professor in the Automatic Control research unit at Chalmers. His research focuses on modeling and automatic control of energy storage systems, particularly lithium-ion batteries. Many of his works are in collaboration with industry partners, such as Volvo Cars, Volvo Trucks, Polestar, Scania, and CEVT AB. He joined Chalmers in 2017 as a Postdoctoral Researcher and became an Assistant Professor in 2019 in the same research unit at Chalmers. He was a visiting researcher at the University of California, Berkeley, USA. He obtained the PhD degree in Automation and Control Engineering from the University of Melbourne, Australia.

16:00-16:30: Monica Marinescu from Imperial college London. Topic TBA.