Madagascar Gecko Optimization Algorithm

Swarm intelligence optimization algorithms solve complex optimization problems by simulating the behavior of groups in nature. However, most existing methods rely on immediate fitness-driven search strategies, which are prone to premature convergence in multi-peak, high-dimensional, or noisy environments. To address this issue, this paper proposes a novel swarm intelligence optimization algorithm—the Madagascar gecko optimization algorithm. Inspired by the attachment, slippage, and detachment behavior of Madagascar geckos in complex surface environments, this algorithm introduces state variables such as attachment memory, orientation confidence, and environmental stability, transforming individual search behavior from "immediate fitness response" to a "state-driven decision-making process." By constructing a temporal convolutional attachment memory kernel, an orientation confidence model based on historical success rates, and a strategic detachment mechanism, the algorithm significantly enhances global exploration capabilities while maintaining local search stability. Furthermore, this paper proposes a non-monotonic search rhythm scheduling method, enabling the algorithm to retain periodic exploration capabilities even in the later stages of iteration. Theoretical analysis shows that this method exhibits good search diversity and stability in multi-peak continuous optimization problems, providing a new approach to state-based modeling of swarm intelligence algorithms.