Learning-Based Scheduling of Flexible Manufacturing Systems Using Support Vector Machines

Dispatching rules are usually applied to dynamically schedule jobs in flexible manufacturing systems (FMSs). Despite their frequent use a significant drawback is that the performance level of the rule is dictated by the current state of the manufacturing system. Because no rule is better than any other for every system state, it would be highly desirable to know which rule is the most appropriate for each given condition. To achieve this goal we propose a scheduling approach using support vector machines (SVMs). By using this technique and by analyzing the earlier performance of the system, “scheduling knowledge” is obtained whereby the right dispatching rule at each particular moment can be determined. Simulation results show that the proposed approach leads to significant performance improvements over existing dispatching rules. In the same way it is also confirmed that SVMs perform better than other traditional machine learning algorithms as the inductive learning when applied to FMS scheduling problem, due to their better generalization capability.