Multi Machine Scheduling: An Agent-Based Approach

Scheduling of multiple parallel machinesin the face of sequence dependent setups and downstream considerations is a hard problem. No single efficient algorithm is guaranteedto produce optimal results. We describe a solution for an instance of this problem, in the domain of papermanufacturing. The problem has additional job machine restrictions and fixed costs of assigning jobs to machines. We consider multiple objectives such as minimizing (weighted) tardiness, minimizing job-machine assignmentcosts. We solve the problem using a simple agent architecture called the Asynchronous team (A-team), in which agents cooperate by exchanging results. We have built agents each of which encapsulatesa different problem solving strategy for solving the multi-machine scheduling problem. The A-team framework enables the agents to cooperate to produce better resultsthan thoseof any individual agent. In this paperwe define the problem, describe the individual agents, and show with experimentalresults t...

• Type: Conference
• Year: 1998
• Where: AAAI
• Authors: Rama Akkiraju, Pinar Keskinocak, Sesh Murthy, Frederick Y. Wu
• Keywords: AAAI 1998, Hard Problem, Intelligent Agents, Multi-machine Scheduling Problem, Problem Solving Strategy