Datum / čas
Date(s) - 13.05.
15:00 - 17:00
Kategorie
Feature-driven Robust Stochastic Scheduling for Printed Circuit Board Assembly
by
Fuzhou University
Abstract:
Motivated by an industry project, this talk addresses scheduling in printed circuit board assembly (PCBA), a bottleneck of electronic manufacturing. We consider uncertain processing and setup times arising from machine variability and human intervention, and model the problem as identical parallel machine scheduling to minimize total completion time and makespan. We develop a feature-driven robust stochastic optimization model that embeds production features into decision-making: processing times are predicted via support vector regression, while setup uncertainty is captured through event-wise ambiguity sets constructed by K-means clustering. The model is reformulated as a mixed-integer linear program and solved using a branch-and-price (B&P) algorithm. Experiments on real-world data show that the proposed approach outperforms sample average approximation (SAA) and standard distributionally robust optimization (DRO) by 52% and 33%, respectively. The B&P algorithm scales well to realistic instances, and sensitivity analysis reveals the impact of setup scenarios on the trade-off between solution quality and computational effort.
Scheduling Seminar
Objective of a virtual seminar on scheduling research and applications is to discuss both the field’s newest advancements and survey traditional areas. Seminars take place typically on every second Wednesday through three different time zones (Europe, the Middle East & Africa, North America & South America, and Asia, Australia & Oceania).