Advanced shop-floor scheduling with genetic algorithm for combined horizon optimization in holonic manufacturing systems

Abstract

Shop-floor scheduling is one of the most complex problems in holonic manufacturing systems as it deals with unpredictable scenarios and overlapping requirements. The scheduling problem is considered to be a NP-hard problem and only near optimal solutions can be obtained. A solution for this problem can be formulated only by employing meta-heuristic class algorithm. This paper discusses the scheduling problem in heterachical operating mode, focusing on solving the local horizon problem. A distributed genetic algorithm is introduced, that uses the local operation plan to generate the initial solution population. The initial population is then evolved based on global optimum soft conditions until the acceptable solution fitness is achieved. The soft conditions considered at global horizon layer are energy footprint, resource utilization and supply chain optimization for resource stocks. The paper describes the data structures used to model this logic and describes in detail the genetic algorithm evolution mechanism. Experimental results are discussed in the context of a pilot production line consisting of six universal resources and a conveyor belt processing two parallel customer orders. © 2013 Springer-Verlag.