Job shop scheduling in manufacturing factories is a problem of finding a sequence of operations to produce given orders. Because of computational complexity, finding the optimal schedule is almost impossible with algorithmic approaches in real situations. Therefore, we often have to be satisfied with a feasible solution that can be sought within the limited resources.
In this thesis, we developed a job shop scheduling tool called PASS (Knowledge-Based Planning and Scheduling System Shell), which consists of a knowledge base, a hierarchical planner, and a window-based user interface. A hierarchical planning strategy is employed in that job shop scheduling problem is divided into high-level planning and low-level scheduling. A rough plan is constructed by the high-level planner, while a detailed schedule is obtained from the low-level scheduler by considering all the constraints. In the case when the low-level scheduler cannot make a schedule to satisfy all the constraints, the high-level plan is amended by confirming with the user. A window-based user interface is developed to include display windows, menu windows, and temporary pop-up windows. Through the use of windows and mouse, a friendly user environment is provided. By successfully developing a job shop scheduling system for a switchboard factory, the usefulness of this tool is assured.
