Feasibility Study of Real-Time Scheduling Using the Lagrangean Relaxation Method Under an APS Environment

摘要

Advanced Planning and Scheduling (APS) has been widely recognized as a promising method for solving real production planning and scheduling problems. Based on the proposal of a real-time job shop scheduling mechanism under an APS environment, which adopts the Lagrangean relaxation method as the optimization logic, the present paper describes a feasibility study of this mechanism by evaluating its calculation speed and re-scheduling quality. Numerical experiments have been carried out for various models having different scales, as well as different densities and strengths of random events, such as the arrival of new jobs or changes to the due dates for existing jobs. The results of experiments show that the proposed scheduling mechanism has the potential to satisfy the real-time scheduling requirements, not only in terms of calculation speed and solution quality, but also with respect to predictability of the calculation load. Finally, an improvement to the Lagrangean relaxation method is proposed to improve re-scheduling quality.