Sequence learning-based schedule prediction for flexible manufacturing systems under uncertainties
Abstract This study presents a method of production schedule prediction for flexible manufacturing systems with consideration of the uncertainty factors including limited machine capacity, diverse processing time and unplanned waiting time. The proposed method can predict product-level schedules using sequence learning, which derives data-learned models to predict production sequence proactively and granularly at the product-level. A decision tree technique is applied to derive such predictive models to pre-trace the locations of individual products allocated to each workstation. A deterministic technique is also applied to estimate waiting and production time per product as well as total production time consumed to fabricate all products assigned by work orders.