Discrete Optimal Design of Trusses by Generalized Extremal Optimization

In this paper the first results of the application of the Generalized Extremal Optimization (GEO) algorithm to a discrete structural optimization problem is shown. GEO is an evolutionary brand new algorithm, devised to be easily applicable to a broad class of nonlinear constrained optimization problems, with the presence of any combination of continuos, discrete and integer variables. So far, it has been applied successfully to real optimal design problems with continuos design variables and shown to be competitive to other stochastic methods such as the Genetic Algorithms (GAs) and the Simulated Annealing (SA). Having only one free parameter to adjust, it can be easily set to give its best performance for a given application. This is an a priori advantage over methods such as the SA and GAs since each of them have at least three parameters to be set, making their tuning to a particular application more prone to be computationally expensive and becoming a problem in itself. In this work, the 10-bar truss problem is used as a test case and the performance of GEO, compared to results from other methods available in literature.