Evaluation of Dispatching Strategies for the Optimization of a Real-World Production Plant

In this paper we present three optimization approaches for a real-world production planning problem, which are based on dispatching. In particular, we employ genetic algorithms to select one rule per machine or machine group from a set of simple dispatching rules and compare the results with a global, composite rule as generated with genetic programming. Reusability and long-term stability of these rules is also investigated on four snapshots extracted from the production planning and control system of our project partner over a time frame of several months.