AI & predictive maintenance of Li-Ion batteries
Description
The presentation addresses the challenges of on-board, stationary and grid energy storage systems, focusing particularly on the limitations of Lithium-ion batteries, such as high cost, limited driving range, slow recharge times, and potential over-costs. It introduces the concept of predictive maintenance, which uses Machine Learning (ML) techniques to estimate the Remaining Useful Life (RUL) of Lithium-Ion batteries. By leveraging data from battery aging tests conducted by research institutes, models are developed to predict both the RUL and the State of Health (SOH) of batteries based on operational data like current, voltage, and temperature readings.
Additionally, the presentation tackles the issue of the explainable AI of predictive models, an area funded by the Horizon Europe ENERGETIC project. This research aims to better understand the relationship between battery usage and its impact on the state of health. It also explores an ontology-based digital passport for batteries, a framework designed to enhance traceability and offer detailed insights into the lifecycle of a battery, including manufacturing details, usage patterns, and degradation metrics. This digital passport is envisioned to promote transparency, support regulatory compliance, and optimize battery recycling and repurposing at the end of its life cycle, contributing to a more sustainable use of battery technology