Optimising the regeneration process of spent lithium‑cobalt oxide cathode through performance analysis model
Authors: Alyoubi, M., Ali, I. and Abdelkader, A.M.
Journal: Journal of Energy Storage
Volume: 110
eISSN: 2352-152X
DOI: 10.1016/j.est.2024.115132
Abstract:The global adoption of renewable energy sources and electric vehicles has substantially impacted the demand for efficient energy storage solutions, specifically lithium-ion batteries (LIBs). Lithium cobalt oxide (LiCoO₂) batteries are widely used for their high energy density and stability. However, the environmental impact and resource depletion associated with the low recycling rate of the exhaust batteries necessitate the development of effective regeneration methods. This study investigates the application of machine learning (ML) models to predict the Remaining Useful Life (RUL) and State of Health (SoH) of regenerated LiCoO₂ cathode, aiming to optimise regeneration conditions and enhance battery efficiency. Using datasets containing specific discharge capacity values across various regeneration temperatures, material ratios, and amounts of lithium and cobalt added to the spent materials, several ML models were developed and compared, including Artificial Neural Networks (ANN), Advance Classification and Regression Trees (C&RT), Boosted Regression Trees (BRT), Support Vector Machine (SVM) and K-Nearest Neighbours (KNN). The accuracy of ANN was unparalleled, attaining an impressive R2 value of 0.993. The model identified an optimal regeneration temperature of 825 °C and a 1–1 ratio with 5.5 mg of n(Li)/n(Co) added for the solid-state regeneration method, significantly improving specific capacity and extending the battery lifecycle. Moreover, the regenerating lithium cobalt oxide at these conditions preserves 98 % of its initial capacity compared to the commercial Lithium cobalt oxide, highlighting its high efficiency and effectiveness at these optimal conditions. Additionally, the model was applied to predict the rate performance under optimal regeneration conditions, yielding results that surpassed those of experimentally determined optimal settings. Based on recent research that utilised the same regeneration method, validation has been achieved by comparing the results of this study with our findings, confirming the predictive precision of the model and emphasising its potential to reduce the reliance on extensive laboratory testing and improve the cost-effectiveness of battery regeneration processes. This research underscores the transformative role of machine learning in advancing sustainable and efficient energy storage technologies.
Source: Scopus