A market approach to decentralized control of a manufacturing cell

Based on a fictitious market model, a decentralized approach is presented for the workstation scheduling in a CNC workshop. A multi-agent framework is proposed, where job agents and resource agents act as buyers and sellers of resource in the virtual market. With cost and benefit calculation of these agent activities, which reflects the state of the production environment, various, and often conflicting goals and interests influencing the scheduling process in practice can be balanced through a unified instrument offered by the markets. The paper first introduces a heuristic procedure that makes scheduling reservations in a periodic manner. A multi-agent framework is then introduced, in which job agents and resource agents seek appropriate job–workstation matches through bidding in the construction of the above periodic “micro-schedules”. A pricing policy is proposed for the price-directed coordination of agent activities in this. Simulation results demonstrate the feasibility of the proposed approach and give some insights on the effects of some decision making parameters. Future work will be focused on the designing of some more sophisticated coordination mechanism and its deployment.