Optimizing engineering design problems using adaptive differential learning teaching-learning-based optimization: Novel approach

As the complexity of mathematical optimization problems intensifies in real-world scenarios, the imperative to devise sophisticated algorithms becomes evident. Consequently, researchers are intensifying their focus on formulating efficient optimization methodologies capable of adeptly navigating the feasible space. This involves enhancing established metaheuristic algorithms through the integration of diverse evolutionary procedures. The main contribution of this paper is development of an adaptive differential learning teaching-learning-based optimization (ADL-TLBO) method for effectively and reliably optimizing unknown parameters in engineering design problems. ADL-TLBO incorporates four enhancements: i) Adaptive selection between the teacher and learner phases of TLBO based on learners' ranking probabilities; ii) Introduction of an adaptive crossover rate to enhance population variety, determined by the learners' rating process; iii) Integration of differential learning (DL) to enable a broader exploration of the search area by learners during the learner phase; iv) Implementation of an accelerator mechanism to expedite convergence during the optimization process. ADL-TLBO is tested on twenty-three test functions and three real-world engineering design challenges to validate its efficiency. Comparisons reveal that ADL-TLBO exhibits superior optimization efficacy compared to other state-of-the-art competitors. ADL-TLBO outperforms other approaches in terms of convergence speed and computational effort, mainly applied to real engineering problems.