Abstract
Solution of complex engineering problems by meta-heuristics, requires powerful operators to maintain sufficient diversification as well as proper intensification during the search. Standard Imperialist Competitive Algorithm, ICA, delays the search intensification by propagating it via a number of artificial empires that compete each other until one concurs the others. An Enhanced Imperialist Competitive Algorithm is developed here by adding an evolutionary operator to the standard ICA followed by greedy replacement; in order to improve its effectiveness. The new operator introduces a walk step directed from the less fit to the fitter individual in each pair of the search agents together with a random scaling and pick up scheme. EICA performance is then compared with ICA as well as GA, PSO, DE, CBO, TLBO, SOS; first in a set of fifteen test functions. Second, a variety of continuous and discrete engineering benchmarks and structural sizing problems are solved to evaluate EICA in constrained optimization. In this regard, a diversity index is traced as well as the other convergence metrics. The results exhibit considerable improvement of the algorithm by the proposed features of EICA and its competitive performance with respect to the other treated methods.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have