An Agent-based Scheduling Method Enabling Rescheduling with Trial-and-Error Approach

摘要

Scheduling optimization is an extremely difficult problem; therefore, many scheduling methods such as linear programming or stochastic searching have been investigated in order to obtain better solutions close to the optimum one. After obtaining a certain solution, scheduling managers may need to reschedule another solution that corresponds to changes in requirements or resources. However, rescheduling problems become more difficult as they become larger in scale. In this paper, we propose an agent-based rescheduling system using the linear programming approach. In our system, agents can autonomously conduct rescheduling on behalf of managers by repeated trial and error in balancing loads or changing the priority of resource allocation until it reaches a better solution for the requirement is obtained. In addition, managers can engage in trial and error with the help of agents to seek a better solution by changing constraint conditions.