Universität Ulm, Fakultät für Informatik, Abtl. Künstliche Intelligenz up: Abteilungsseminar KI

Constraint Satisfaction in Complex and Dynamic Scheduling Problems

Prof. Dr. Roman Barták
Department of Theoretical Computer Science and Mathematical Logic,
Charles University, Prague

01.12.2003



 
 Abstract
The talk is devoted to techniques of applying constraint satisfaction technology to solve complex and dynamic scheduling problems. First, we argue for a more tighten integration of planning and scheduling, some examples from real-life manufacturing problems are given. A short introduction to constraint satisfaction technology, which can be used to solve such problems, follows. The main part of the talk is divided into two components. First, we describe techniques developed to model and solve complex scheduling problems in manufacturing with emphasis to resources with transition schemes and arbitrary resource dependencies. These techniques are used in the advanced planning and scheduling system Visopt ShopFloor. Second, we describe a new approach to dynamic constraint satisfaction problems where the number of changes in the solution should be kept minimal after changes in the problem - a so called minimal perturbation problem (MPP). Ideas of a solving algorithm for MPP are sketched and preliminary results in timetabling are presented. The talk is concluded by case studies from food and pharmaceutical industries.
Abtl. KI Startseite Hilfe Mail an Webmaster B. Schattenberg - 24.11.2003