Enhancing population diversity based gaining-sharing knowledge based algorithm for global optimization and engineering design problems

Abstract

Gaining-sharing knowledge based algorithm (GSK) is a recently emerged meta-heuristic algorithm based on human behavior and has been successfully applied to solve various optimization problems. However, GSK tends to get trapped in local optimum due to the rapid loss of population diversity during the optimization process, resulting in an imbalance between exploration and exploitation. To overcome this deficiency, this paper proposes an enhancing population diversity based GSK (EPD-GSK) framework. The proposed EPD-GSK framework incorporates three components: (1) The utilization of Sobol sequence with low divergence to initialize the population, enhancing the diversity of initial solutions. (2) The integration of the Cauchy mutation strategy in the junior phase to perturb individuals and expand the search space. (3) The application of the reverse learning update mechanism in the senior phase, increasing the likelihood of escaping local optimum. These techniques promote population diversity throughout the exploration and exploitation stages. The proposed EPD-GSK framework was evaluated on CEC2017, CEC2020, and the latest CEC2022 test suites as well as on four constrained real-world engineering design problems. The experimental results demonstrate that EPD-GSK can effectively improve the performance of various existing GSK algorithms. Furthermore, EPD-GSK also exhibits better performance compared with other state-of-the-art algorithms.