This work improves and applies the adaptive particle swarm optimization algorithm to the study of statistical model fitting of atmospheric turbulence profiles. First, an improved adaptive particle swarm optimization algorithm is proposed to improve the speed of particle swarm optimization and avoid falling into the local optimum. The distance between the current particle and the global optimal position is used to adjust the inertia weight coefficient and make nonlinear adaptive changes. The self-learning and social learning factors are based on the concept of symmetrical linear change to realize the adaptive change of the optimization focus in each stage. Second, the improved adaptive particle swarm optimization algorithm is introduced to solve the generalized Hufnagel-Valley turbulence model in Ali region, and the turbulence model profiles of morning, evening, and four seasons in the region are fitted. The simulation results show that all the decision coefficients are greater than 0.997, which agrees well with those of the statistical average profiles obtained by radiosonde. The proposed method has similar convergence accuracy to other adaptive particle swarm optimization algorithms, but the speed is faster. This paper introduces a new method for fitting Hufnagel-Valley turbulence profile models.