Abstract in English:
The aim of this thesis is to design and evaluate an ant algorithm for a certain class of (hybrid) flowshop problems.
These problems consist of a two-staged production process with two identical machines on the second station and buffers in between the stages. For a given number of jobs, each consisting of a processing time and a deadline, the goal is to create a feasible partition and permutation of these jobs which minimizes the total tardiness. Furthermore, a theoretical investigation on the placement and size of the buffers and its impact on the resulting solution space was done.
In order to further improve the swarm intelligence it was combined with a set of different heuristics and pheromone evaluation rules. The resulting variations of the algorithm were later tested and evaluated on a set of benchmark instances. Subsequently, the obtained theoretical insights on the flowshop problems were analyzed in a practical matter and proved to be valuable. Finally, the quality of the solutions found by the designed ant algorithm was analyzed on smaller pro- blem instances. It became apparent that at least for such smaller problems the algorithm creates very good solutions which only differ little from the optimum.