Blind Cavefish Optimization Algorithm

To address the problems of traditional swarm intelligence algorithms in high-dimensional complex optimization problems, such as susceptibility to local optima, high similarity of convergence structures, and insufficient theoretical analysis, this paper proposes a novel swarm intelligence optimization method—the Blind Cave Fish Optimization Algorithm. Inspired by the lateral line perception, path memory, and cave community structure of blind cave fish, the algorithm constructs an asymmetric lateral line tensor field, an information entropy-driven step size adjustment mechanism, an adaptive cave topology structure, reversible and irreversible dual-mode dynamics, and an exponentially decaying memory kernel function. This method does not directly rely on gradient information but instead forms an anisotropic update mechanism through a direction-selective perturbation field and utilizes local entropy measurement to adaptively adjust the exploration intensity. This paper presents the complete mathematical model, dynamic expressions, and energy function construction, and provides a theoretical discussion of the algorithm's boundedness and asymptotic stability. Results show that this model is structurally significantly different from classical particle swarm optimization, differential evolution, and gray wolf algorithms, exhibiting strong high-dimensional adaptability and the ability to escape local optima.