Machine Learning for Predicting Battery Health and Remaining Useful Life: Advancing Sustainable Practices in a Circular Economy
Students & Supervisors
Student Authors
Supervisors
Abstract
The rapid growth in demand for batteries is driven by developments in electric automobiles, renewable energy systems, and practical consumer electronics complicated societal and ethical problems related to mining rare materials such as lithium, cobalt, and nickel. Although these components are vital to the life of a battery, eliminating them often causes severe suffering and creates ethical concerns that are worth addressing. Addressing these challenges necessitates a transition to a circular economy with a focus on reusing and recycling batteries. In this work, we focus on creating machine learning (ML) models that monitor the battery status and remaining useful life (RUL), which may aid in cost-effective battery management solutions. Using a dataset of 14 NMC-LCO 18650 batteries that have lasted 1000 cycles, we classify batteries into health categories such as "good life," "needs attention," and "end of life" using classification models, XGBoost and Random Forest. This activity is necessary to accurately detect the battery state and predict the remaining useful life (RUL) to improve battery efficiency, enable second-life applications, and manage the reuse process. Envisioned benefits include time-delayed assessment of battery life, forecasting which batteries should be reused or repurposed, devising sustainable reuse systems, etc. This idea brings up the potential of machine learning approaches to establish a circular economy, reduce biological feedback, and boost the maintainability of energy storage devices.
Keywords
Publication Details
- Type of Publication: Conference
- Conference Name: 3rd National Mathematics Conference 2024 by the Department of Mathematics, BUET
- Date of Conference: 06/02/2025 - 07/02/2025
- Venue: Department of Mathematics, BUET
- Organizer: Department of Mathematics, BUET