Refinement of Dynamic Hunting Leadership Algorithm for Enhanced Numerical Optimization

Abstract

A recently created optimization algorithm named the Dynamic Hunting Leadership (DHL) algorithm was inspired by the leadership tactics used in hunting operations. The foundation of DHL is the idea that successful leadership can significantly increase hunting endeavors. Although DHL has shown to be relatively simple and successful in tackling a variety of practical optimization issues, it has been discovered it suffers with efficiently balancing global exploration and local search phase, particularly in high-dimensional numerical problems and engineering applications. Furthermore, due to drawbacks, it is vulnerable to becoming stuck in local optimal. The present study aims to tackle the aforementioned challenges by introducing a modified variant of DHL, referred to as mDHL, that utilizes the Levy Flight technique as a localized development strategy to augment each hunter’s capacity to track their prey and attain superior optimal outcomes. Moreover, local escape operator and quasi-opposition learning are synergistically incorporated to foster the hunters’ exploration and localized optimal escape techniques. These tactics foster superior knowledge sharing between leaders and hunters, resulting in a harmonious blend of exploration and development capabilities. The mDHL algorithm is shown to outperform existing optimizers across 20 function test suites with varying dimensions from 30 to 200 and CEC 2019 functions. In addition, it has been successfully applied to solve four practical engineering design cases, demonstrating its practicality. The experimental findings indicate a substantial improvement over conventional DHL, emphasizing the potential of mDHL as a competitive and efficient algorithm for addressing engineering and numerical optimization challenges.

Keywords

• Convergence acceleration
• engineering constrained design problem
• metaheuristic algorithm