The simulated annealing method is introduced into genetic algorithm to form a hybrid algorithm, which can overcome the defects of genetic algorithm in optimization and avoid the premature phenomenon. According to the approximate formula of reflection coefficient of P-SV wave, the objective function of anisotropy parameter inversion is established. The anisotropy parameter inversion of TTI medium is achieved by using single algorithm and mixed algorithm respectively. The results show that the mixed algorithm inversion accuracy is higher and the process is more stable. Furthermore, Gaussian white noise with a signal-to-noise ratio of 5 is added to the data of the theoretical model to test the anti-noise property. The error of the inversion results is small and the anti-noise ability is good. Finally, the anisotropic parameter inversion of the modified Hess model is carried out, and the inversion profile is obtained, which is in good agreement with the original profile, and the effectiveness of the hybrid algorithm is verified. In addition, the research area is selected to perform inversion test on the actual data. The inversion results are good and the accuracy is high, which verifies the effectiveness of the hybrid algorithm in practical application.