Developments of the cluster oriented genetic algorithm

The paper reviews the development of the cluster-oriented genetic algorithm (COGA) strategy from the initial approach to more recent advances which significantly improve the performance of COGA in both the search capabilities and the consistency of the generated design solutions. COGAs are specifically designed to identify high-performance (HP) regions of complex, multi-variable design spaces whilst also achieving good set cover in terms of solutions across these regions. The objective is to extract information from such regions relating to the nature of the problem space in addition to providing the designer with a succinct collection of HP design options. The application of COGA to a number of real-world design tasks is discussed, and also its integration within a graphical user interface and the interactive evolutionary design system.