Sensitivity analysis of fuzzy-genetic approach applied to cabled-truss design

Abstract

This paper sheds light on a fuzzy-genetic system applied to optimize cabled-truss structures. The optimization procedure combines ground structure approach, nonlinear finite element analysis, genetic algorithm, and fuzzy logic. The latter is used to include expertise in the evolutionary search to classify and filter individuals with low survival possibility (s p ). The classification is based on a scale that varies from 0% to 100%, and the filtering depends on a threshold value (S pt ) defined by the user. Particularly, the individuals with s p ≤ S pt are not evaluated, thereby decreasing the total number of evaluations. Although this approach proved suitable to reduce computational cost, the effect of different S pt values on the system's performance was not yet investigated. In that light, this work aims to present a sensitivity analysis of the fuzzy-genetic optimization system to variations of S pt . For that, the system was applied to ground structures with 10 elements and S pt values ranging from 0% to 100%. The results were compared by means of the analysis of variance test in order to investigate the effects of S pt on the system's performance and to identify the optimum S pt , which was found to be of 60% for the studied case.