This research addresses a novel Flexible Manufacturing System focused on rescheduling production which combines the techniques of Preti Nets(PetN), discrete simulation and memetic algorithms such as genetic algorithms(AG), simulated annealing (SA) and local search (LS). Reactive scheduling is caused by a failure of the machines. This hybrid algorithm is called as "PetNMA" and consider the performance measures such as makespan and Total Weighted Tardiness (TWT). The reactive program considers the operations which have not yet been processed at the time of the failure and the operation is processed by the machine at that moment. The PetNMA algorithm consists of three sub-algorithms: the first for the initial scheduling; the second to simulate the execution of the initial program set until the machine failure occurs; and the third to build reactive scheduling through the application of a Memetic Algorithm. We consider the problems that represent an FMS those are contrasted with bottleneck heuristics and dispatch rules obtaining good results in the proposed objectives. In the initial schedule instances, the proposed genetic algorithm improves in 50%, 70% and 70%, in the makespan, total weighted tardiness and combination respectively. The 58% of the results in rescheduling have a variation between 0% and 2% relative to the initial scheduling. The 95%, 85% and 85% of the rescheduling results show a variation between 0% and 2% with respect to the dispatch rules for the Makespan, the total weighted tardiness and its combination respectively.
Reactive scheduling of multiple EOSs under cloud uncertainties: Model and algorithms
