Harris Hawk Optimization (HHO) demonstrated superior performance, high efficiency and efficient computation in various engineering and global optimization problems. However, this algorithm still has various limitations and constraints, including slow convergence, population diversity problems and stagnation of local optima. To overcome these shortcomings, this study introduces an Adaptive Harris Hawk Optimization (ADHHO) for solving complex global optimization problems.
Four important improvements are proposed to the original HHO. Initially, chaotic maps are used for population initialization and generating random numbers. Secondly, a nonlinear control parameter is employed to keep a balance between global and local search capabilities. Thirdly, an adaptive inertia weight (AIW) strategy is introduced to improve convergence speed and stability. Finally, an adaptive decay factor is applied to the Lévy flight, which helps control the steps toward final convergence. To test and validate the success of the proposed ADHHO, two popular benchmark suites are used, CEC2005 and CEC2014. The results on these benchmark suites are then evaluated against five advanced metaheuristic algorithms. Additionally, the performance of the ADHHO is further compared with recently proposed HHO variants and a hybrid combination of HHO and other advanced algorithms.
The experimental results show that the ADHHO converges faster and presents the lowest final fitness values, particularly for complex global optimization problems. The proposed ADHHO achieved the theoretical best solution on 31 out of 40 metric values for CEC2005 and 38 out of 44 metric values for the CEC2014 benchmark suite. Similarly, the ADHHO outperforms all studied HHO variants, achieving the best scores on 28 out of 40 metrics. Overall, the proposed ADHHO offers high-quality solutions to real-world optimization problems that are crucial in the fields of renewable energy, engineering design and computer science.
This research proposes an adaptive HHO, an efficient tool for solving global optimization problems. The proposed algorithm, combined with chaotic maps, a nonlinear energy factor, an AIW and an adaptive decay factor, effectively addresses the limitations of the original HHO and optimizes the final fitness value, convergence rate and efficiency in complex global optimization problems.
