中文摘要：

对于大型洞室群工程，基于三维数值模型进行力学参数识别具有重要意义，但三维数值模拟正算过程需消耗大量时间，且解容易限于局部最小，传统优化方法受到限制。为解决上述问题，本文在已有研究基础上，引入支持向量机和遗传算法，提出了一种新的力学参数识别方法——基于三维数值模拟-SVM非线性模型的洞室围岩参数进化识别方法。将该方法应用于清江水布垭电站地下厂房工程的围岩参数识别，获得了令人满意的效果。

英文摘要：

It is useful to identify the mechanics parameters based on three dimensional numerical model for large cavern group project, but the three dimensional numerical simulation expends too much time and the solution is easily confined to local least, so that the traditional optimistic methods are limited. Based on the former research, the support vector machine and GA is adopted, a new method of rock mechanics parameters identification——the evolution identification method based on three dimensional numerical simulation of nonlinear model is presented in this paper. The method is used for rock mechanical parameters identification of the underground powerhouse of Shuibuya project, the result is satisfactory.