A novel multi-objective improved teaching-learning-based algorithm combined with local search method
Zhang Mei, Yang Shengxuan · 2017
For the purpose of solving multi-objective optimization problems and obtaining good convergence and well-spread Pareto-optimal front, we propose a novel multi-objective improved teaching-learning-based algorithm combined with local search method (MO-ITLLS), which adopts the framework of NSGA-II. In MO-ITLLS, the new candidate population is created by performing the improved teaching-learning algorithm combined with local search method (ITLLS). ITLLS consists of three major phases, namely preschool phase, teacher phase and learner phase. Different with basic TLBO, a preschool phase is designed in ITLLS to increase the diversity of population and to enhance the exploration capability. A divide-and-teaching strategy combined with local search method is proposed in teacher phase to improve the exploitation capability. Furthermore, a modified learner strategy with random dimensions permutation is presented in learner phase to further strengthen the exploitation capability. Finally, the proposed algorithm is tested on UF problems, and it is compared with state-of-the-art algorithms including NSGA-II, SPEA2, MOEA/D, OMOPSO, MOTLBO, and HTL-MOPSO. Experimental results validate the effectiveness of MO-ITLLS and demonstrate its superior capability to find well-spread solutions of better convergence.