To address the challenges of low efficiency and heavy dependence on operator expertise in existing fiber alignment and coupling techniques, this study introduces an fiber automatic alignment coupling system that uses a spiral scanning algorithm and a dual adaptive segmented stochastic parallel gradient descent (DAS SPGD) algorithm. Coarse alignment is conducted using the spiral scanning algorithm to narrow the search scope, and fine alignment is achieved using the DAS SPGD algorithm. A fast and stable alignment process is achieved by adaptively adjusting the gain and perturbation amplitudes. The simulation and experimental results indicate that the DAS SPGD algorithm outperforms other algorithms in terms of convergence rate, stability, and reliability. Notably, in scenarios involving significant radial migration, the effective convergence range of the DAS SPGD algorithm is increased by 5 times compared with other algorithms, highlighting its practical effectiveness. These findings provide valuable insights for advancements in fiber optic communications technology.