Author(s)

J.E. Bolander, Jr, Y. Kobashi & J. Kintzel

Abstract

Genetic algorithms are applied to maximizing the stiffness of structural members represented by highly indeterminate truss networks. Truss ele- ment cross-sectional areas are the design variables to be optimized. Unlike conventional algorithms which use a static set of solution parameters, the number of design variables is reduced during the evolutionary process. Deci- sions to contract the design variable space are based on surveys of the design population, where the functionality of each genetic component is taken into consideration. Upon reducing the number of variables, a new population is initialized with random values and the optimization continues. This re- generation process dramatically improves performance, both in spee

Keywords