Eventos Anais de eventos
COBEM 2023
27th International Congress of Mechanical Engineering
Data-driven model to estimate State-of-Health (SoH) from Lithium-Ion Batteries
Submission Author:
Carlos Antônio Rufino Júnior , MG , Brazil
Co-Authors:
Carlos Antônio Rufino Júnior, Hans-Georg Schweiger
Presenter: Carlos Antônio Rufino Júnior
doi://10.26678/ABCM.COBEM2023.COB2023-2104
Abstract
The breakdown of a battery is a potentially catastrophic occurrence that may have severe repercussions. The ability to forecast when a Li-ion battery would die is currently quite restricted. The primary contribution that this study makes to the field is an investigation into the feasibility of determining whether or not thermal runaway signs may be efficiently identified in advance of the commencement of the phenomenon by making use of currently available measuring methods for heat, temperature, acoustic emission, Coulomb efficiency, or electrochemical processes. The application of these experiments to detect fault occurrences (such as short circuits) in aged cells (also known as second-life batteries) is another one of the new aspects of this body of work. Another innovation that might come out of this study is the determination of the most precarious state of battery cells—that is, the state in which a mechanical failure or a short circuit is most likely to take place. Probabilistic models are a viable option to consider in this scenario. Therefore, the next critical step is to determine the predicted size of these signals of interest, which are significant for commencing harmful trajectories, and then measure them. This will be done before moving on to the following crucial phase. The most significant obstacle is the fact that cells with the same shape, chemistry, and history may (or may not) display distinct mistakes when subjected to the same stimuli (mechanical, electrical, and thermal). Collecting trustworthy data that can give physical insights into events that cause battery failures is the purpose of the proposed thesis. Additionally, the collected data will be used to feed data-based models that are able to make predictions about battery failures (for example, an internal short circuit). These models are used in order to cut down on the amount of actual tests, and as a result, save both money and time.
Keywords
Electric vehicle battery, Artificial Intelligence, Artificial neural networks, Deep learning
