Publication: Optimization of autonomous mobile robots (AMRS) task in semiconductor manufacturing production line
Loading...
Date
2024-07
Authors
Chen, Jun Xian
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Over the past decades, the field of vehicle routing optimization has witnessed significant advancements, driven by the demands of modern industries. As manufacturing processes evolve towards greater automation, connectivity, and data-driven decision-making, the optimization of vehicle routing plays a crucial role in enhancing operational efficiency and productivity within supply chains. Hence, a General Variable Neighbourhood Search algorithm is developed to further optimise the Autonomous Mobile Robots (AMRs) task in semiconductor manufacturing production line, aiming to minimize both the total distance travelled and computational time for generate solutions. This algorithm integrates saving heuristics, shaking phase, and the Variable Neighbourhood Descent or local search, taking advantage of their complementary strengths to improve solution quality and computational efficiency. The developed approach aims to find near-optimal solutions within a reasonable computational time by utilizing the power of saving heuristics as initial solution, which identify and combine routes to achieve distance or time savings. The GVNS then refines this initial solution through systematic changes in the neighbourhood structure, effectively exploring the solution space and avoiding local optima. By integrating these methods, the approach ensures both efficiency and robustness, making it suitable for solving large-scale Vehicle Routing Problems with Time Windows (VRPTW) and various other logistical challenges. Furthermore, the use of the exact method in this work helps to validate the results acquired by the GVNS algorithm. The proposed GVNS algorithm's accuracy and reliability can be assessed by comparing the solutions created by the algorithm to those produced by the exact algorithm, providing confidence in its effectiveness. The algorithm also improved 15 out of 30 (50%)
instances from Solomon Benchmark, demonstrating its effectiveness.