Genetic Algorithm Based Approach for the Multi-Hoist Design and Scheduling Problem

Abstract

Electroplating facilities often face the Cyclic Hoist Scheduling Problem when a repetitive sequence of moves is searched for the hoists. This paper addresses this optimization problem extended to the design of the workshop, where we aim to minimize both the cycle time and the number of hoists used. For this goal, we propose a genetic meta-heuristic approach which introduces a novel solution encoding to enlarge the solutions' search space. Our encoding procedure is based on hoists' empty moves, and includes separator characters. With the latter, we obtain solutions that were not reachable by previous approaches. Each solution obtained thanks to the genetic operators is evaluated by using a Mixed Integer Linear Program. This one checks the constraints of the problem (such as capacity constraints and soaking time bounds) and computes the smallest cycle time for a given moving sequence and its associated number of hoists. Some results are presented using benchmark instances for which our approach allows to improve the best known solutions.