Deadlock-free scheduling of automated manufacturing systems using petri nets and hybrid heuristic search

This paper focuses on the deadlock-free scheduling problem of automated manufacturing systems with shared resources and route flexibility, and develops novel scheduling methods by combining deadlock control policies and hybrid heuristic search. Place-timed Petri nets are used to model the systems and find a feasible sequence of firing transitions in the built model such that the firing time of its last transition is as small as possible. Based on the reachability graph of the net and a minimum processing time matrix, new heuristic and selection functions are designed to guide search processes. The proposed hybrid heuristic search is based on state space exploration and hence suffers from the state space explosion problem. In order to reduce the explored space, the search is restrained within a limited local search window. By embedding the deadlock-avoidance policies into the search processes, a novel deadlock-free hybrid heuristic search algorithm is developed. Experimental results indicate the effectiveness and superiority of the proposed algorithm over the state-of-the-art method.