A Tabu-Based Exploratory Evolutionary Algorithm for Multiobjective Optimization

This paper presents an exploratorymultiobjective evolutionary algorithm (EMOEA)that integrates the features of tabu search andevolutionary algorithm for multiobjective (MO)optimization. The method incorporates the taburestriction in individual examination andpreservation in order to maintain the searchdiversity in evolutionary MO optimization,which subsequently helps to prevent the searchfrom trapping in local optima as well as topromote the evolution towards the globaltrade-offs concurrently. In addition, a newlateral interference is presented in the paperto distribute nondominated individuals alongthe discovered Pareto-front uniformly. Unlikemany niching or sharing methods, the lateralinterference can be performed without the needof parameter settings and can be flexiblyapplied in either the parameter or objectivedomain. The features of the proposed algorithmare examined based upon three benchmarkproblems. Experimental results show that EMOEAperforms well in searching and distributingnondominated solutions along the trade-offsuniformly, and offers a competitive behavior toescape from local optima in a noisyenvironment.