Assessment of mathematical programming and agent-based modelling for off-line scheduling: application to energy aware manufacturing

Abstract State-of-the-art approaches to energy aware scheduling can be centralized or decentralized, predictive or reactive, and they can use methods ranging from mathematical programming to agent-based distributed models. In this paper two methodologies are compared for off-line scheduling, for energy intensive manufacturing systems by using a real industrial case, specifically manufacturing by injection moulding. A multi-objective scheduling problem requiring the minimization of the jobs tardiness and energy consumption is faced. A mixed integer programming formulation and a multi-agent approach are proposed and evaluated. Their advantages and drawbacks are pointed out for off-line energy-aware scheduling.