Hybridization of the imperialist competitive algorithm and local search with application to ship design optimization

In the wide scenario of the optimization techniques, a large number of algorithms are inspired by natural processes, in many different ways. One of the latest is the Imperialist Competitive Algorithm (ICA) Atashpaz-Gargari and Lucas (2007), judged by their authors as very efficient and competitive with other popular optimization algorithms. However, its diffusion is still limited, so that it has not yet been adequately studied. In this paper, we have investigated the convergence properties of the ICA algorithm, observing the effect of the various coefficients and their role in the global convergence. Some modifications, including the coupling with a local search method, have been listed/suggested and then tested on a suite of standard algebraic test functions, verifying the improvements on the speed of convergence of the original algorithm. An application to naval design has been also included, in order to check the ability to solve realistic problems.