Résumé
In electric vehicle applications, operating conditions heavily affect the battery cell lifetime and cost. The aging process of Lithium-ion Battery (LiB) cells is influenced by numerous interrelated stress factors, making it challenging to predict aging levels accurately and develop effective mitigation strategies. Including machine learning (ML) models in the Battery Management System (BMS) enables real time analysis and informed decision-making process based on multi-factorial data. To limit degradation of LiB cells, it is crucial to scrutinize the effect of each stress factor and explore interdependencies. We propose an ML model designed to classify aging cycling/storing conditions and protocols based on LiB cell capacity degradation. The proposed model is intended for implementation in the BMS and distinguishes itself through training on industrial datasets. The consideration of large number of features and their interactions with a balanced dataset results in greater interpretability of our model. We further advance the model by developing an interactive application leveraging ML outcomes, allowing for real time navigation over the space of parameters.