Imperialist competitive algorithm for design optimization of crane metallic Structure

Abstract

In design optimization of crane metallic structures (CMSs), existing heuristic approaches require very high computational cost for searching global or near-global optimal solutions. This hindered their generalization and application in practice. In this paper a novel discrete imperialist competitive algorithm (ICA) for the design optimization of CMSs is developed and firstly used for a real crane metallic structure. For comparison of performances, a genetic algorithm, particle swarm optimization with mutation local search, ant colony algorithm, ant colony algorithm with mutation local search and the proposed ICA are together used to optimize the same structure. The optimization results show that the computational cost of ICA is approximately 4.0% of that of the GA, around 2.6% of that of the PSOM, and about 4.8% of that of the ACA. The objective function value given by ICA is 23.68% less than the practical design value, a reduction of 16.42% over the GA, 18.84 % over the PSOM and 15.5 6 % over the ACA. These illustrated that the developed ICA is an effective and efficient intelligent method for design optimization of CMSs and is superior to other well-known heuristic algorithms.