Scheduling of ̄exible manufacturing systems: an ant colony optimization approach

The scheduling problem for ̄exible manufacturing systems (FMSs) has been attempted in this paper using the ant colony optimization (ACO) technique. Since the operation of a job in FMSs can be performed on more than one machine, the scheduling of the FMS is considered as a computationally hard problem. Ant algorithms are based on the foraging behaviour of real ants. The article deals with the ant algorithm with certain modi®cations that make it suitable for application to the required problem. The proposed solution procedure applies a graph-based representation technique with nodes and arcs representing operation and transfer from one stage of processing to the other. Individual ants move from the initial node to the ®nal node through all nodes desired to be visited. The solution of the algorithm is a collective outcome of the solution found by all the ants. The pheromone trail is updated after all the ants have found out their respective solutions. Various features like stagnation avoidance and prevention from quick convergence have been incorporated in the proposed algorithm so that the near-optimal solution is obtained for the FMS scheduling problem, which is considered as a non-polynomial (NP)-hard problem. The algorithm stabilizes to the solution in considerably lesser computational e€ ort. Extensive computational experiments have been carried out to study the in ̄uence of various parameters on the system performance.