Application of evolutionary optimization methods to design of space trusses

This paper presents an application of evolutionary optimization methods to design of space trusses. The objective is to minimize the weight of space trusses with discrete variable sections subject to constraints on displacements, stresses, and local stability. Member's displacement sensitivity index, stress index and slenderness index are derived and used to change the member section. The process of structural analysis - member indexes calculation - member section selection is repeated until all constraints are satisfied. At each iteration, a number of members can only be assigned the next smaller or larger sections from an initially chosen set of sections. This discrete section sizing approach provides much convenience in practical design works. Various optimization algorithms, based on section selection strategy, type and sequence of constraints included into the algorithm, are proposed and employed. The optimization process is automatically carried out by an optimization program called FEMOPT. Optimal designs obtained by the proposed algorithms are very similar to each other and can be compared with solutions by other optimization methods.