Distributed scheduling based on multi-agent systems and optimization methods

Abstract

The increasing relevance of complex systems in dynamic environments has received special attention during the last decade from the researchers. Such systems need to satisfy products or clients desires, which, after accomplished might change, becoming a very dynamic situation. Currently, decentralized approaches could assist in the automation of dynamic scheduling, based on the distribution of control functions over a swarm network of decision-making entities. Distributed scheduling, in an automatic manner, can be answered by a service coordination architecture of the different schedule components. However, it is necessary to introduce the control layer in the solution, encapsulating an intelligent service that merge agents with optimization methods. Multi-agent systems (MAS) can be combined with several optimization methods to extract the best of the two worlds: the intelligent control, cooperation and autonomy provided by MAS solutions and the optimum offered by optimization methods. The proposal intends to test the intelligent management of the schedule composition quality, in two case studies namely, manufacturing and home health care.