Dynamic scheduling of flexible manufacturing system using support vector machines

Abstract

A flexible manufacturing system (FMS) needs a powerful scheduler to assign dispatching rules dynamically for achieving good performance. A scheduler should possess high generalization ability to tackle unpredictable conditions such as different part types, part mix ratios, and job arrivals. This paper presents a support vector scheduler, which is based on the support vector machine (SVM), to achieve the goal of dynamical scheduling. SVM is superior to other traditional learning machines such as multilayer neural networks for the FMS scheduling because it possesses better generalization performance. To justify the simulation results, the well-known FMS model and physical layout widely used in the FMS scheduling are employed in this paper. Using support vector scheduler combined with the kernel of radial basis function (RBF), simulation results show that the throughput performance is better than the one using static dispatching rules. In addition, the design process of the SVM-based scheduler for the FMS model was accomplished in a very short time. Therefore, it can be fast implemented for other different FMSs to achieve the optimal performance.