A robust meta-hyper-heuristic approach to hybrid flow-shop scheduling

Abstract

Combining meta-heuristics and specialised methods is a common strategy to generate effective heuristics. The inconvenience of this practice, however, is that, often, the resulting hybrids are ineffective on related problems. Moreover, frequently, a high cost must be paid to develop such methods. To overcome these limitations, the idea of using a hyper-heuristic to generate information to assist a meta-heuristic, is explored. The devised approach is tested on the Hybrid Flow Shop (HFS) scheduling problem in 8 different forms, each with a different objective function. Computational results suggest that this approach is effective on all 8 problems considered. Its performance is also comparable to that of specialised methods for HFS with a particular objective function. © Springer-Verlag Berlin Heidelberg 2007.