Therefore, this paper proposed a mixed genetic algorithm and simulated annealing (GASA) algorithm for polarization control. Based on a genetic algorithm (GA), this algorithm introduced the selection mechanism of simulated annealing (SA), and introduces the Monte Carlo idea into GA, which made the algorithm have more powerful search ability. By establishing a mathematical model for the active polarization control system of fiber laser based on the GASA algorithm, we obtained the simulation images of the GASA algorithm under different parameter combinations and analyzed the convergence effect. The simulation results show that when the polarization extinction ratio of output laser is selected as the fitness function, the population number is 90, the mutation probability is 0.7, the crossover probability is 0.001, and the temperature drop ratio is 0.99, the system can achieve the optimal control effect. By comparing the simulation images of GASA algorithm and stochastic parallel gradient descent algorithm, we can see that the GASA algorithm has better ability of global search and jumping out of local optimal values, and can be used in the active polarization control system of fiber laser.