To address the problems of isotropic diffusion, directional oscillation, and dimensional redundancy inherent in traditional swarm intelligence optimization algorithms in high-dimensional complex spaces, this paper proposes a swarm optimization method based on anisotropic wave topology and irreversible energy dynamics. The algorithm achieves asymmetric information diffusion by constructing a direction-dependent propagation kernel; captures local curvature structures through a direction matrix self-learning mechanism; builds direction memory through irreversible energy dynamics; suppresses repeated searches through wave interference cancellation; achieves structural dimensionality reduction through submanifold locking; and uses vibrational spectrum energy as the structural convergence criterion. This method models the swarm search behavior as a topologically driven discrete dynamical system, re-characterizing the swarm optimization process from the perspective of information propagation structure and directional energy flow. Theoretical analysis shows that under appropriate parameter conditions, the system possesses a stable attractive domain and has the ability to balance global search and local convergence.