A GA-based decision support system for professional course scheduling at Ford Service Organisation

Abstract.

We describe the concept and implementation of a decision support system for professional course scheduling which allows the planner to generate, evaluate, and compare different schedules obtained by different runs based on variations of the objective function and different strategies of blocking, i.e. pre-assigning certain subsets of courses. The core of the system is a population-based improvement heuristic, i.e. a specific genetic algorithm. We have implemented the decision support system as web-based service, i.e. a distributed client server system with a thin client, where communication between server and client is via the Internet. The approach has been developed for the specific planning situation at Ford Service Organisation in Germany and has shown to significantly improve the planning process with respect to quality of schedules, time-to-plan, and flexibility.