MULTICRITERIA OPTIMIZATION OF AIRCRAFT PANELS - DETERMINING VIABLE GENETIC ALGORITHM CONFIGURATIONS

This article compares different Genetic Algorithm (GA) configurations for multicriteria optimization using the design of helicopter flat panels as a test bed. The results show that an efficient set of GA configurations can be determined through a two-step process and that the techniques used produced results as optimal as provided by expert rule-based techniques but found them in 1/100 the time. Phase 1 of the experiment established and tested a set of GA baseline configurations; Phase 2 expanded the most optimal baseline results into a set of preferred configurations. Over 700 GA configurations were tested and the effects of the different configurations are compared to previous studies based on single criterion optimization. Guidelines for identifying and testing GA configurations are summarized. (C) 1995 John Wiley & Sons, Inc.