Data-Driven Decision-Support for Maintenance Operations in Stacking Battery Production

Abstract

In the contemporary landscape of battery production, the integration of data-driven predictive maintenance strategies stands out as a pivotal enhancement to operational efficiency and equipment reliability. This study focuses on the predictive maintenance of the stacking process in lithium-ion battery (LiB) production, a critical phase with a significant impact on total productivity of the organisation, due to frequent downtimes associated with the stacking cutting module. Utilizing a CRISP-DM framework within the mixed methods research methodology , this thesis examines the practical implementation of predictive maintenance by analyzing historical data and conducting qualitative research through workshops and literature review. The research highlights the critical role of data quality and availability in the successful application of machine learning models for enabling predictive maintenance. Key findings suggest that while the potential for data-driven maintenance to improve operational efficiency is substantial, significant challenges remain in data collection, system integration, and aligning technological advancements with organizational goals. This study contributes to the body of knowledge by outlining a structured approach in implementing predictive maintenance in battery manufacturing and providing insights into the technical considerations necessary for leveraging Industry 4.0 technologies effectively. The thesis supports industry practitioners in transitioning towards data-driven decision-support for enabling predictive maintenance models, thereby enhancing the sustainability and competitiveness of battery production facilities.