Group Discussion Based Dynamic and Adaptive Strategies for Enhancing Teaching–Learning-Based Optimization

The Teaching–Learning-Based Optimization (TLBO) algorithm simulates the pedagogical dynamics of the classroom to guide population-based search. While TLBO exhibits rapid convergence through effective exploitation of the search space, its enhanced variant—Group TLBO (GTLBO)—introduces a group discussion mechanism during the learning phase to promote diversity and mitigate premature convergence. However, empirical analysis reveals that in early iterations, when selected solutions for discussion are widely dispersed, the algorithm may generate excessively large steps. This can lead to premature exploitation and insufficient exploration, resulting in suboptimal search behavior. To address this limitation, this work proposes two novel variants: Dynamic Group TLBO (DGTLBO) and Adaptive Group TLBO (AGTLBO), which incorporate dynamic and adaptive scaling of learning steps, respectively. Their constrained counterparts, HSR-DGTLBO and HSR-AGTLBO, are further developed by integrating with them a Hybrid Stochastic Ranking (HSR) based constraint-handling mechanism. The proposed algorithms are evaluated on the CEC 2017 benchmark suites and a set of four engineering design problems. The comparative analysis of extensive simulation results demonstrates that adaptive variants outperform dynamic variants, and both are superior to their parent variants and other state-of-the-art algorithms on the tested benchmarks and engineering problems. Additionally, an interaction group size of three yields optimal performance across all the tested cases.