Work

Civet Optimization Algorithm: A Novel Metaheuristic Optimization Approach for Global Numerical Optimization

Year: 2026
Type: article
Abstract: Metaheuristic optimization algorithms are widely used to solve complex optimization problems; however, designing new mechanisms to strike an appropriate balance between exploration and exploitation is crucial.In this study, the Civet Optimization Algorithm (COA) is introduced as a novel algorithm inspired by the natural behaviour of the Asian palm civet (Paradoxurus hermaphroditus).The core concept of COA is based on the relationship between the animal's movement and home-range utilization, which are associated with food resource availability, and its foodselection behaviour.Based on this, two primary behaviours are selected to design the search mechanisms.In the exploration phase, "resource-dependent movement and spatial displacement" are modeled; as the relative quality of the current position decreases, the algorithm models greater displacement and exploration of new areas of the search space.In the exploitation phase, "local residence and selective feeding" behaviours are employed, focusing the search on promising regions by selecting the nearest superior food location.These two behaviours form the COA framework through complementary mathematical models.The performance of COA is evaluated using 29 benchmark functions from the CEC 2017 test suite-comprising unimodal (C17-F1 and C17-F3), multimodal (C17-F4-F10), hybrid (C17-F11-F20), and composition (C17-F21-F30) functions-and compared against nine other metaheuristic optimization algorithms.The results demonstrate that COA ranks first in 25 out of the 29 functions, accounting for 86.21% of the benchmark functions.Furthermore, boxplot analysis provides additional evidence regarding the distribution and stability of COA's results across independent runs.The findings indicate that COA, by striking an appropriate balance between exploration and exploitation, supports the potential applicability of COA to a broader range of optimization problems.The development of binary and multi-objective versions, investigation of discrete and constrained optimization problems, and application of COA to real-world problems represent important directions for future research.
Language: English
Open Access status: hybrid
OpenAlex ID: W7234842845
FWCI:
Cites: 31
Cited by:
Related to: 10