Real-Time Routing in Flexible Flow Shops:

In this talk a self-adaptive, swarm-based control model for real-time part routing in a flexible flow shop environment is presented.. The proposed control model is a multi-agent system that exhibits adaptive behaviour, which has been inspired from the natural system of the wasp colony. The production problem, which was first introduced by Morley et al. (Morley and Schelberg 1993, Morley 1996, Morley and Ekberg 1998), consists of assigning trucks to paint booths in real-time in a flexible flow shop environment with the objective of throughput maximization and minimization of number of paint flushes accrued by the production system, assuming no a priori knowledge of the colour sequence or colour distribution of trucks is available. The proposed control model is benchmarked with the results of Cicirello and Smith (2004) in solving the same production problem. The proposed control model uses self-adapting threshold parameters, as opposed to that of Cicirello and Smith’s (2004) hand-tuned (i.e., based on trial and error) parameters. A simulation-based software environment is designed and developed to investigate its performance. The simulation results show that the proposed control model is more robust to environmental changes and it outperforms the previous models proposed by Morley (1996) and Cicirello and Smith (2004) on the basis of throughput, number of setups, and cycle time performance measure.