Ensembles of Heuristics and Computational Optimisation in Highly Flexible Manufacturing System

The objective of a Flexible Manufacturing System (FMS) is to respond faster to changes in products and demands with minimum changeover cost. However, layout changes in FMS are not automatic and required human intervention. Therefore, when requirements for layout changes are frequent, such as in a dynamic production environment, like mass personalisation production environments, layout reconfiguration becomes expensive and unrealistic. In this paper, we relax this core assumption of static FMS layout and introduce a decentralised approach to the design and coordination of manufacturing systems’ entities, whereby both products and production machines are mobile and autonomous. We apply three different optimisation methods, of which two are ensembles of computational and heuristics optimisation approaches based on Gradient Descent and Ant Colony Optimisation (ACO), to optimise mobile machines locations under non-deterministic manufacturing conditions as obtainable in a mass personalisation context. These approaches enable mobile production machines to coordinate and autonomously adjust their location and layout in real-time to minimise the cost of material flow between production machines. The proposed approach offers a promising outlook on the design and coordination of manufacturing systems under unpredictable manufacturing conditions.