介绍了用遗传算法在三维横向不均匀介质中确定震源位置的研究思路并简述了三维横向不均匀介质中走时计算的基本原理．在遗传算法的迭代过程中，考虑到初始迭代阶段有较大的参数搜索空间而在最后迭代阶段有较小的参数空间，提出了遗传算法中变异概率随迭代次数下降的函数形式并对其进行了分析．结果表明，随迭代次数指数下降的变异概率使得算法收敛加快，而且反演得到的震源位置的各台站走时残差比原定震源位置的明显减小 The research idea of ​​determining the position of hypocenter in three - dimensional inhomogeneous media by using genetic algorithm is introduced and the basic principle of travel time calculation in three - dimensional transverse inhomogeneous media is briefly introduced. In the iterative process of genetic algorithm, considering that there is a large parameter search space in the initial iteration stage and a smaller parameter space in the last iteration stage, a function form of the mutation probability decreasing with the iteration number in genetic algorithm is proposed and carried out Analysis. The results show that the algorithm converges faster with the mutation probability of decreasing index number of iteration, and the travel time residual of each station at the source position retrieved by the inversion is obviously smaller than that of the original hypocenters