An improved variant of teaching-learning- based optimization (TLBO) known as constrained teachinglearning- based optimization with modified learning phases (CTLBO-MLPs) is proposed in this paper to tackle the constrained optimization problems more effectively. A constraint handling technique is first incorporated into the CTLBO-MLPs in order to ensure the feasibility of solutions. Significant modifications are then proposed to refine the TLBO’s teaching and learning framework in order to achieve better balancing of the exploration and exploitation searches in the proposed algorithm. Particularly, a modified teacher phase is introduced to maintain the population diversity by assuming each learner is guided by their unique perceptions of mainstream knowledge represented through mean position. Moreover, the learner phase of CTLBO-MLPs is enhanced via an adaptive peer learning scheme and a self-learning scheme. The former scheme enables a CTLBO-MLPs learner to improve the knowledge by interacting with multiple learners in each dimensional components, while the latter one simulates an alternate knowledge improvement strategy with personal efforts. The performance of CTLBO-MLPs in solving constrained optimization problems is evaluated using the CEC 2006 benchmark function set and compared with eight peer algorithms. Extensive simulation studies show that the proposed CTLBO-MLPs has the best search performance among all compared algorithms.
Volume 12 | 04-Special Issue
Pages: 1442-1456
DOI: 10.5373/JARDCS/V12SP4/20201623