An agent-based dynamic scheduling for parallel machine under machine eligibility constraints

Abstract

This research applies agent-based negotiation approach to construct a multi-agent-based framework and solve parallel machines rescheduling with eligibility. The proposed architecture comprises three types of agents: job agents, machine agents, and supervisor agent. The decision model is built in each job/machine agent for bid selection to ensure better global objectives can be reached. The supervisor agent is used to manage negotiations between job agents and machine agents and monitor the production process. When there are some disruptive events occurred, the supervisor agent will arrange dynamically a reactive scheduling to update the current production scheduling in order to provide an immediate response to the unexpected event. The objectives are multiple, which include total weighted tardiness, total flow time, and utilization.