Hybrid Algorithm of Adhesive Joint Shape Optimization

The subject of this article is the algorithm for optimization of symmetrical adhesive lapped joint. The goal of the paper is development of the method for optimization problem solving, which allows to unite high rate of calculations with stability of obtained results. This goal can be reached by means of two algorithms of optimization – genetic algorithms implemented on the original stage and swarm of particles algorithm – on the last stage of optimization. The problem of optimization is in finding optimal shape of doublers, i.e. doubler length and function of thickness variation along joint length. To describe stress-strain state of a joint modified Holland-Reissner model is used. To solve direct problem of structural stress state estimation the finite elements method is used. For optimization problem solving combination of multi-population model of genetic algorithm and swarm of particles algorithm are used. Introduction of individuals from other populations to the considered one allows to escape of homogenization of genotypes in separate population and premature breakage of optimization process. To describe doubler shape thickness variation function development of Fourier series are used. Above-mentioned implemented methods allow to create algorithm for topologic optimization which unites advantages of both methods and find solution of considered problem quite quickly. Duration of algorithm realization on Python language requires several minutes only to find optimal parameters.