Near-optimal scheduling of line production with an evolutionary algorithm

The paper presents an evolutionary scheduling algorithm designed for and applied to a group of production lines in an automobile factory. The cost of schedules is determined by the quality of energy consumption management over peak demand periods, and the task is to minimize the energy costs by appropriate scheduling of process interruptions. To explore the capabilities of the evolutionary scheduling algorithm, its results are compared with optimal solutions produced for selected problem instances by a constraint logic programming (CLP) approach. The comparison shows the evolutionary algorithm is capable of finding near-optimal schedules while successfully handling a larger set of possible problem instances than CLP.