The Research of Hybrid Scheduling Model in AS/RS Based on Multi-agent System and Petri Net

This paper studies the problem of loading/unloading scheduling in AR/SR (Automated Storage/Retrieval System) and establishes a hybrid scheduling model which takes advantage of pure centralized and distributed structure. The scheduling model based on multi-Agent and Petri Net is proposed, and the structure of the physical layer, control layer and management layer of the model is detailed described. The distributed sub Agent in the bottom level uses AHP (analytic hierarchy process) to assign weights to the scheduling rules and also uses Genetic Algorithm and Petri Net to optimize the rules. The centralized main Agent in the upper level mainly uses improved Hungarian Algorithm to assign tasks. Loading/unloading scheduling in AR/SR is controlled by making use of collaborative communication of the Agents. JADE is used as the development platform to resolve tasks in/output decision problems of AS/RS. It shows that the hybrid scheduling model is effective according to the simulation example.