围岩介质多具有非线性，而对其进行开挖施工实质上是一个不断对围岩进行反复加卸载的复杂过程。因此，这时围岩的稳定就不仅是和最终状态有关，而是与施工过程有关，或是与应力路径或应力历史有关，所以正确合理的选择分期分步开挖方案是非常必要性的。而数值分析方法的研究与进展，大大地促进了地下结构稳定型评价和工程设计的定量化进程。 与此同时，为了选取合理的开挖方案，必定涉及到施工顺序的优化问题。大型地下工程分步开挖顺序的可能组合非常多，而遗传算法具有较强搜索功能和进行全局优化的特点，因此运用遗传算法来进行优化搜索。在该文研究中，把数值方法和遗传算法进行结合，充分发挥数值方法的数值计算能力和遗传算法的全局搜索能力，从而给出了从可能的地下工程施工顺序中选取最优方案的方法。在进行遗传算法优化的实现过程中，针对地下工程开挖的实际情况，对编码方案进行了研究，采用符号编码法进行编码；从围岩的破坏机理出发及考虑问题实现的实际情况，提出了以开挖后形成的塑性面积作为目标函数，以此来确定遗传算法中的适应度函数的方法，并设计了适合的遗传算子；针对遗传算法处理地下工程开挖顺序优化问题时出现的特殊情况，提出了自己的看法和解决途径。编制了能确定出优化的开挖顺序的程序。并以该程序分析了工程实例，分析结果表明了文中方法及所研制程序的正确性和有效性，其具有一定的工程应用研究价值。

Most of wall rock media are nonlinear substance，and the construction of excavation for them is materially a complex process during which surrounding rock is installed and uninstall repeatedly. Thus, the stabilization of wall rock not only relates to final state, but also to the process or stress path or stress history, and it is necessary to select reasonable by-stage excavation scheme. The investigation and development of numerical analyze method out and away promote quantification of stabilization estimate in underground configuration and design of project. In order to select reasonable project of excavation, optimum gradation of construction is come down to. The likelihood combination of gradation in big scale subsection excavation is very much, and Genetic Algorithms have better searching function, and then Genetic Algorithms can optimize object as a whole, therefore, Genetic Algorithms can be used as a optimum searching method. The optimum project can be selected from likelihood construction project in underground engineering through combining numerical method and Genetic Algorithms, which adequately exert the calculation function of numerical method and the optimum searching function of Genetic Algorithms. In the course realization of Genetic Algorithms optimum, according practical instance of underground engineering excavation, the coding method is investigated and symbol-coding method is bought forward for this instance; according mechanism of wall rock destroying and considering the practical instance, the plastic area is considered as object function formed after excavation, and then fitness function of Genetic Algorithms is confirmed; the Genetic Algorithms operators is designed; in the course of dealing with optimum design of excavation order in underground engineering, some special condition may occur. My opinions and some resolving approach are brought forward. The program was made to confirm optimum design of excavation order in underground engineering. Using this program a engineering example has been analyzed, which indicates this program can guide practical engineering.