Improvement of Artificial Hummingbird Algorithm with Greedy Fitness-Distance Balance Method and Its Application to...

This paper proposes an improved Artificial Hummingbird Algorithm (AHA) for solving numerical optimization problems. The greedy Fitness-Distance Balance (FDB) selection method is integrated into the guided, territorial, and migration search operators of the AHA. This establishes a balance between exploration and exploitation by enhancing AHA's exploration capabilities. The performance of the proposed FDB-AHA is evaluated using CEC 2017 and 2020 problems which include single-mode, multi-mode, hybrid, and composition functions in search spaces of dimensions 30, 50, and 100. Furthermore, the results of the FDB-AHA are compared with 11 other advanced search algorithms using the Friedman statistical test method on the CEC 2017 and 2020 problems. As a result of these comparisons, the FDB-AHA was found to be the most effective method, while the AHA ranked ninth. These findings confirm the effectiveness of the improvements made in the AHA. Moreover, the FDB-AHA has been used to determine the optimal solutions for load frequency control in three different power systems. In this context, three-term (PID), five-term (FOPID), and six-term (FOPI-FOPD) controllers are optimized and compared with existing studies in the literature for time-domain analysis. Furthermore, load variation and robustness analysis are also performed. Finally, the practicality of the controllers optimized by FDB-AHA is verified using the real-time simulator OP5707. Simulation results show that FDB-AHA is a highly efficient algorithm for solving all optimization problems.