Designing Cable-Stayed Bridges with Genetic Algorithms

Authors

Abstract

Cable-stayed bridges construction involves the determination of a high number of design variables, both static and dynamic. Moreover, the properties of such variables make them statically indeterminate, meaning that a change in one design variable affects the response of the entire structure. This property makes the design of cable-stayed bridges a complex optimization problem. In this work, we use a Genetic Algorithm to evolve solutions for this problem. A set of experiments are executed, where conventional variation operators are used for exploring the solution space. The first experiments suggest that this is a problem with a deceptive landscape. However, we show that we can design solutions that optimize structural objectives. Moreover, we want also to minimize costs while presenting different optimized solutions. In the second set of experiments, we included a baseline solution in the population to evaluate if we could find better solutions using this approach. The results on the second set showed that it was possible, thus we moved to the third set of experiments with more parameter tuning. The experimental results suggest that we are able to find new and suitable solutions for the problem comparable to the existing baseline approach.

Conference

Applications of Evolutionary Computation 2020