Self-adaptive Wolf Pack Algorithm based on Dynamic Population Updating for Continuous Optimization Problems

Wolf pack algorithm (WPA) is a relatively new swarm intelligence-based algorithm for solving complex continuous optimisation problems as well as real-world optimisation problems. The basic WPA and its variants are prone to trap into local optima and premature convergence when tackling multi-modal functions due to diversity loss problem and imbalance between exploration and exploitation. Inspired by the idea of integrating the heuristic information and stochastic strategies to balance exploration with exploitation, we propose a self-adaptive WPA based on dynamic population updating (SWPA-DU) strategy. First, the self-adaptive chaotic scouting behaviour is designed to develop the global exploration of scout wolves. Second, a novel Cauchy perturbation operator is proposed to generate a few mutation besieging wolves, which not only enhances the capability of jumping out of local optima but also improves local exploitation. Third, a dynamic population updating strategy is invented to improve diversity. Numerical experiments with a suit of benchmark functions and practical applications are performed to verify the effectiveness and advancement of the proposed algorithm. The experimental results indicate that SWPA-DU obtains superior performance on both multi-modal and high-dimensional problems over the compared algorithms.