A decomposition based algorithm for the scheduling problem in wafer fabrication system

Abstract

A decomposition based scheduling strategy is proposed to cope with the complexity of wafer fabrication, such as multi-products, reentrant process flow and unrelated parallel machines. The strategy is composed of genetic algorithm and dynamic scheduling rules. First, bottlenecks are detected and classified based on the constraint weights which are decided by the sensitivity of the system performance to the machine's fluctuant availability. Second, a dynamic work in process (WIP) control strategy is used to determine the dispatching order by adjusting the deviation of the WIP levels to avoid the process fluctuation caused by unpredictable events. Third, a hybrid genetic algorithm (GA) is used to optimize the dispatching route of the lots. Finally, a simulation platform is developed to validate the effectiveness of the proposed method.