조선에서 조립공장간 통합 스케줄링 시스템의 개발

Abstract

The production process of shipbuilding is very complex in nature. To build an efficient shipbuilding scheduling system so as to reduce the inherent computational complexity, DAS (Daewoo Shipbuilding Scheduling System) has adopted a two-layered hierarchical architecture. In the hierarchical architecture, individual scheduling systems composed of a higher-level dock scheduler, DAS-ERECT and lower-level assembly plant schedulers, DAS-PBS, DAS-3DS, DAS-NPS, and DAS-A7 try to search the best schedules under their own constraints. However, as an individual scheduler does not communicate with others during scheduling an assembly plant, another similar individual scheduler can not resolve trivial problems caused by an individual scheduler. That is, even if an individual scheduler optimizes an assembly plant, it guarantees only the local optimum. In this paper, we suggest Knowledge-Based Load Balancing and Load Transfer Algorithm to resolve the problems among the multiple assembly plant schedulers. The one is a 3-dimensional look ahead load balancing before scheduling assembly plants using the domain knowledge, while the other is a 3-dimensional post load transfer after scheduling assembly plants. A prototype scheduling coordination system, DAS-ASICS (Assembly Schedule Integrated Control System), is implemented as a subsystem of DAS using the expert system development tool UNIK on SUN SPARC station. A series of experiments using the real-world data are also carried out to guide the equilibrium of load among assembly plant schedulers.