Abstract: By use of 8 complex functions, a heterogeneous multi group particle swarm optimization algorithm is simulated, comparing with the traditional single population particle swarm optimization algorithm. Aiming at the deficiencies of the stage discharge relation fitting in it is difficult to determine the parameters, using a variety of the heterogeneous particle swarm optimization algorithm for optimizing the relative parameters of the stage-discharge relationship, taking stage-discharge relation fitting of Yunnan Province Longtan station, Xiyang station as case studies, and the particle swarm optimization algorithm and least squares fitting results are compared in the study. The analysis results show that the convergence accuracy of the heterogeneous multi group particle swarm optimization algorithm is much better than the particle swarm optimization algorithm,with good computational robustness and global optimization ability. The relative error absolute values of the fitting for the relationships between the water level of the Longtan railway station and the Xiyang station are only 0.27% and 0.50% respectively. The fitting accuracy is better than that of the particle swarm optimization and the least square method. The better fitting effect can be obtained by optimizing the water level flow relationship by using the heterogeneous multi group particle swarm optimization algorithm.