Augmented Harris hawks optimization and for engineering design problems and UAV path planning

LD Zhu and YF Fu, INTERNATIONAL JOURNAL OF MACHINE LEARNING AND CYBERNETICS, 16, 6295-6345 (2025).

DOI: 10.1007/s13042-025-02624-x

Harris hawk optimization (HHO) is a popular metaheuristic algorithm recently proposed, but it suffers from slow convergence speed, low accuracy, and the problem of easily getting stuck in local optima when dealing with practical problems. To address these drawbacks, we propose an improved version called augmented Harris hawk optimization (AHHO). In the enhancement, we introduce Logistic chaotic population initialization to enhance AHHO's global exploration capability; adopt a dynamic L & eacute;vy flight strategy to improve the algorithm's early convergence speed; and propose a new surprise pounce exploitation strategy to enhance the algorithm's optimization ability. Additionally, we introduce a random centroid dynamic backward learning to improve the algorithm's search efficiency, convergence speed, and robustness, thereby effectively solving complex optimization problems. To validate the performance of AHHO, we conducted analyses from three aspects: population diversity, exploration and exploitation balance, and convergence behavior. We compared it with HHO and 12 other high- performance algorithms on the CEC2014 and CEC2017 test sets, in two test sets, 15, 17, 23, 14, 18, and 21 first-place rankings were obtained in three dimensions, respectively. The results show that compared to HHO and other algorithms, AHHO demonstrates superior comprehensive performance in all aspects. Finally, we applied AHHO to 5 engineering optimization problems and 1 unmanned aerial vehicle path planning problem, 5 engineering problems and UAVs both achieved first place rankings. The results indicate that AHHO outperforms other benchmark algorithms, demonstrating its strong scalability and outstanding optimization performance.

Return to Publications page