A Review of Dynamic Scheduling: Context, Techniques and Prospects

Quite a lot of literature exists for static scheduling for shop-floor; static schedules, however, become obsolete almost immediately as the systems experience unpredictable disruptions on an almost non-stop basis. Dynamic scheduling techniques have thus become the key techniques to achieve robust and flexible production schedules. Researchers have proposed multiple approaches to handle the unpredictable disruptions on the shop floor, either by predictively developing a robust and stable schedule or designing an efficient reactive policy to assign resources to tasks. These approaches are different from each other in context, complexity, and effectiveness. In this paper, we review the state of the art of research into dynamic scheduling for production systems including heuristics, meta-heuristics, multi-agent system (MAS) and machine learning (ML) based techniques. We also review popular techniques of scheduling in various contexts including flow shops, job shops and their variants, paying special attention to their adaptation in dynamic environments. We conclude with a discussion of opportunities and challenges that would arise in adopting the machine learning based techniques to schedule production resources in face of exploding real-time shop-floor data that accompanies the emergence of 4th industrial revolution.