Confined teaching-learning-based optimization with variable search strategies for continuous optimization

Abstract

The well-known optimization approach teaching-learning-based optimization (TLBO) is modified by using a confined TLBO (CTLBO) to eliminate the teaching factor. Different settings are suggested for various types of search factors, as they are used for different purposes. In addition, crossover frequencies are introduced into TLBO to prevent premature convergence. Furthermore, eight new mutation strategies are introduced to the teacher phase, and four new mutation strategies to the student phase to enhance the algorithm's exploitation and exploration capabilities. The experimental results show that the proposed versions, especially those that either adopted low crossover frequencies or implemented various mutation strategies, performed particularly well in achieving fast convergence speeds in the early stages, reaching convergence precision at lower cost, arriving at convergence plateaus at either lower cost or higher precision, handling tests of composition functions well, and achieving competitive performance on CEC2015 test problems.