King Cobra Algorithm (KCA) and its application in engineering optimization

Abstract

Engineering optimization problems require algorithms that are accurate, computationally efficient, and adaptable to diverse search conditions. This paper proposes the King Cobra Algorithm (KCA), a population-based metaheuristic inspired by the king cobra's hunting behavior. KCA combines two movement strategies: prey-seeking exploration and focused-attack exploitation, controlled by a phase-selection mechanism that balances global and local search. The experimental study employs a multi-stage design. Initially, internal sensitivity analyses on the CEC-2014 benchmark suite assess the Gaussian-function component, population size, and phase-selection mechanism. Subsequently, KCA is compared on CEC-2014 with CEC/WCCI competition winners and top-ranked optimizers under a fixed maximum computation time, enabling a practical, runtime-oriented comparison with competition-proven methods. Finally, KCA is evaluated on the CEC-2017 benchmark suite under a fixed number of objective-function evaluations and compared with representative standard metaheuristic optimizers. Statistical interpretation is conducted using Friedman rankings, critical-difference diagrams, Holm post-hoc tests, and paired Wilcoxon signed-rank tests. Two variants, the simple KCA (sKCA) and the self-adaptive KCA (jKCA), are introduced to investigate fixed and adaptive phase-selection rules. The KCA-based methods are subsequently applied to chemical and power engineering optimization problems, such as economic load dispatch and optimal power flow. Results indicate that KCA delivers competitive benchmark performance and high-quality feasible engineering solutions, while sKCA and jKCA illustrate the flexibility of the proposed framework.