A Petri net-based particle swarm optimization approach for scheduling deadlock-prone flexible manufacturing systems

This paper proposes an effective hybrid particle swarm optimization (hpso) algorithm to solve the deadlock-free scheduling problem of flexible manufacturing systems (fmss) that are characterized with lot sizes, resource capacities, and routing flexibility. based on the timed petri net model of fms, a random-key based solution representation is designed to encode the routing and sequencing information of a schedule into one particle. for the existence of deadlocks, most of the particles cannot be directly decoded to a feasible schedule. therefore, a deadlock controller is applied in the decoding scheme to amend deadlock-prone schedules into feasible ones. moreover, two improvement strategies, the particle normalization and the simulated annealing based local search, are designed and incorporated into particle swarm optimization algorithm to enhance the searching ability. the proposed hpso is tested on a set of fms examples, showing its superiority over existing algorithms in terms of both solution quality and robustness.