A Hybrid Deep Learning Framework for Predictive Maintenance in Lithium-Ion Battery Manufacturing

Abstract

Lithium-ion battery production is an important industry as there is a growing demand for energy storage in areas such as transportation, consumer electronics, energy generation, and other industrial applications. Continuous manufacturing operations demand that the equipment used for production be of high quality in order to ensure manufacturing efficiency and continuity. Equipment breakdowns and machine malfunctions often interfere with the manufacturing operations, causing maintenance costs, delay in production, poor-quality production, and lack of equipment availability. Traditional approaches to maintenance involve the use of set maintenance plans or maintenance upon breakdowns and malfunctions of the machinery involved, which lead to ineffective maintenance of equipment and failure in fault detection. Intelligent Predictive Maintenance approach can help detect equipment degradation before the critical point so that proper planning for maintenance can be done. This work aims to develop an approach for predictive maintenance in lithium-ion battery manufacturing through the use of a hybrid deep learning model trained by operational data captured by multiple sensors in the manufacturing process. We introduce a new dataset named Lithium-Ion Battery Manufacturing Predictive Maintenance Dataset (LIBMPMD-2026). The proposed methodology uses spatial feature extraction, temporal degradation learning, and feature weighting to precisely determine the health condition of equipment and predict maintenance needs in varied environmental settings. There is extensive pre-processing that increases the data quality by normalizing, noise filtering, feature extraction, dimensionality reduction, and class balance. The experiment results indicate considerable gains in predictive accuracy, reliability, robustness, and maintenance decisions leading to lesser downtimes and production disruptions. Health evaluation of the equipment makes it possible to schedule maintenance, increase manufacturing stability, enhance efficiency, increase resource efficiency, strengthen sustainable industrialization, adopt technologies, innovate, and increase manufacturing reliability.