Towards learning pallets applied in pull control job-open shop problem

The current paper studies the concept of learning pallets following the autonomy paradigm; in a Conwip control job-shop/ open-shop system. To realize learning capability for pallets several advantages and methodologies can be employed. Among them are the privileges of closed-loops in Conwip system as well as application of evolutionary intelligence for inspiring learning. Specifically, some features of genetic algorithm (GA) can be used to produce new alternatives and avoid local traps in a decentralized approach, though the GA is a global search method. In addition, fuzzy inference system is employed to distinguish the dynamisms of each station as well as of the entire system, concerning vagueness in real time information, and uncertainty in processing sequence and times. It is shown here that learning pallets (Lpallets) are presenting better records in terms of some criteria, e.g., makespan.