Harp Seal Optimization Algorithm Based on a Novel Selection–Combination Technique

Abstract

Metaheuristics are widely used to address optimization problems, but their efficacy varies considerably across different problem instances. This performance variability stems mainly from poor balance between exploration and exploitation. To address this limitation, this paper introduces the Harp seal (HaS) optimizer, inspired by the natural behaviors of HaSs. HaS consists of three procedures: novel selection-combination, migration, and pupping. The proposed work aims to efficiently solve a wider range of complex optimization problems. To validate the effectiveness of HaS, it was tested on 19 classical benchmarks, 10 CEC-2019 benchmarks, and a real-world application. The achieved results compared to well-known algorithms, including the genetic algorithm, multi-verse optimizer, and learner performance-based behavior. HaS outperformed or equaled other algorithms in 14 out of 19 classical benchmarks and 6 out of 10 CEC-2019 functions. It has shown robust capability across unimodal, multimodal, and composite benchmarks. The results demonstrate the proposed algorithm’s capacity for both exploration and exploitation. Moreover, the good trade-off between exploration and exploitation enhances the ability of the algorithm in optimizing large-scale optimization problems. Furthermore, statistical analysis verified the significance of the observed improvements. Overall, HaS offers robust and superior results that have outperformed existing state-of-the-art algorithms.