MOEICA: Enhanced Multi-Objective Optimization based on Imperialist Competitive Algorithm

In this paper, a multi-objective enhanced imperialist competitive algorithm (MOEICA) is presented. The main structures of the original ICA are employed while some novel approaches are also developed. Other than the non-dominated sorting and crowding distance methods which are used as the main tools for comparing and ranking solutions, an auxiliary comparison approach called fuzzy possession is also incorporated. This new provision enables more countries to participate in guiding the population towards different searching routs. Moreover the computational burden of the algorithm is abated by carrying out the hefty sorting process not at each iteration but at some predefined intervals. The frequency of which is controlled by on optional parameter. Furthermore, the recreation of empires and imperialists several times during the optimization progress, encourages better exploration and less chance to get trapped in local optima. The eligibility of the algorithm is tested on fifteen benchmark functions in terms of different performance metrics. The results through the comparison with NSGA-II and MOPSO shows that the MOEICA is a more effective and reliable multiobjective solver with being able to largely cover the true Pareto fronts (PFs) for the test functions applied in this article.