INVESTIGATION OF REVOLUTION OPERATOR IN IMPERIALIST COMPETITIVE ALGORITHM (ICA)

This paper reviews the effects of revolution operator in Imperialist Competitive Algorithm (ICA). The Imperialist Competitive Algorithm that was recently introduced and this algorithm has shown its good performance in optimization problems. This novel optimization algorithm is inspired by socio-political process of imperialistic competition in the real world. This algorithm converges fast but in high dimensional problems, this algorithm easily stuck into local optima. The strategy of this algorithm for escaping from local optima is revolution operator. The revolution is an operator which changes the position of a spot. The revolution operator can be applied in N different form for a N-Dimension problem. These N different form include revolution with 1, 2, ... and N dimension and the results of each one is different from others. In this paper, the effect of dimension's number in revolution operator has been reviewed in Imperialist Competitive Algorithm for multi-objective optimization problems. The revolution operator in basic algorithm choice randomly one of variables and changes its value randomly in each iteration. In this paper, ICA algorithm has been tested under some well-known benchmarks and with comparison of results, the best method for revolution is introduced in order to increasing performance of the ICA.