Finding worst-case flexible schedules using coevolution

Finding flexible schedules is important to industry, since in many environments changes such as machine breakdowns or the appearance of new jobs can happen at short notice. In this paper a minimax formulation is used to develop a coevolutionary algorithm for finding worst case flexible schedules. A population of schedules is used to locate the schedule with the best worst case performance, while a population of breakdowns is used to locate the worst breakdown and estimate the performance of the schedules. This approach is compared to a standard scheduling approach and concluded to produce more flexible schedules. It is also compared to an approach in which the schedules are tested against all possible breakdowns; the coevolutionary approach is found to be faster and produce schedules of a comparable quality.