Scheduling AMSs with generalized Petri nets and highly informed heuristic search

Abstract

The design of the heuristic function in a Petri-net(PN)-based A${}^{\ast }$ search significantly impacts search efficiency and schedule quality for automated manufacturing systems (AMSs). In Luo et al. (2015), two admissible heuristic functions were formulated for an A${}^{\ast }$ search based on place-timed PNs to schedule AMSs. To broaden its application scenarios and enhance search efficiency, this paper proposes a new heuristic function whose calculations take account of multiple resource acquisitions, weighted arcs, redundant resource units, and outdated resources, which are commonly encountered in practical AMSs but usually not considered. The proposed one can deal with generalized PNs, offering broader application scenarios than ordinary PNs. In addition, it is proven to be admissible and more informed than its counterparts, ensuring that the obtained schedules are optimal and making the timed PN-based A${}^{\ast }$ search more efficient. To validate the efficacy and efficiency of the proposed method, several benchmark systems are tested.