RELICH Marcin University of Zielona Gora

Spoluautoři WITKOWSKI Krzysztof, SANIUK Sebastian, SUJANOVA Jana

Decision making supported by task-oriented software tools plays more and more role in production enterprises, including metallurgical plants. Decision making in production planning and scheduling requires the response in an interactive real-time mode. It opens up for a new generation of decision support system (DSS) that enables a fast prototyping of production flows in multi-project environment. The paper aims at providing a knowledge base approach allowing one to be independent of context or representation data as well as allowing for the design of an interactive and task-oriented DSS. The assumed knowledge base mode of specifying a production system leads to solving a decision problem formulated in terms of constraint satisfaction problem (CSP). Possible ways of the CSP decomposition as well as possibility of different programming languages application lead to a problem of searching for a distribution strategy which gives a possibility to interact in a real-time mode. A declarative form of the description of a multicriteria decision problem allows its implementation in constraint programming languages and facilitates the development of a DSS. Illustrative example concerns optimal steelmaking process scheduling with constraints such as processing time, limited waiting time between adjacent stages, and release/due time. Numerical experiments show that constraint programming approach, under appropriate formulation and search strategies, can not only describe the problem exactly, but also can solve the problem more effectively and efficiently compared with classical algorithms.