Genetic algorithm for minimizing tardiness in flow-shop scheduling

The present paper reports on a new approach to applying a genetic algorithm to the flow-shop scheduling problem. Three different objective functions considered are: minimizing total tardiness; minimizing number of tardy jobs; and minimizing both the above objective functions simultaneously. Two sets of solutions are presented; the first is based on a traditional heuristic, the second on a genetic algorithm metaheuristic. The former is suitable for relatively small-scale problem instances, while the latter finds very high quality optimum or near-optimum solution within a reasonably fast time, and is both effective and efficient for both medium- to large-scale problem instances. Results of computational testing are presented and confirm that the approach reported here is of high quality, fast for large problem instances, effective and efficient for flow-shop scheduling.