Akademik Veri Yönetim Sistemi
V. International BRITISH Congress on Interdisciplinary Scientific Research & Practices, London, İngiltere, 5 - 07 Şubat 2026, ss.12, (Özet Bildiri)
In today's industrial production lines, unplanned equipment failures disrupt production continuity, resulting in significant time and cost losses. Predictive maintenance approaches, developed to address these issues, aim to determine the optimal time for maintenance by analyzing sensor data integrated into machines to anticipate potential failures. However, aligning the predicted maintenance time with the production schedule is critically important for ensuring the sustainability of production. This study proposes a comprehensive and dynamic approach by integrating Remaining Useful Life (RUL)-based predictive maintenance systems with a production planning–oriented decision support mechanism. In this context, sensor data obtained from NASA's C-MAPSS dataset are processed using the bi directional long-short term memory (Bi-LSTM) algorithm to estimate equipment-specific RUL values. These estimations are then integrated with a simulated production schedule, taking into account factors such as shift hours, planned downtimes, and production intensity, to propose suitable maintenance time windows. In doing so, a dynamic decision support system that enables real-time synchronization is developed. The performance of the proposed model is analyzed in terms of both RUL prediction accuracy and its compatibility with the production plan. This integrated structure aims not only to ensure technical precision in failure prediction but also to optimize maintenance scheduling by considering operational efficiency. As such, the study presents a dynamic and applicable solution that contributes to the literature as an alternative to conventional static maintenance models.