改进粒子群算法在地下工程反分析中的运用

Application of modified particle swarm optimization to back analysis of underground engineering

结合粒子群算法和变尺度法的特点,充分考虑二者的互补性,提出一种改进粒子群优化算法,并将其运用于地下工程增量位移反分析中.根据围岩松动圈的形成特点,提出采用分区分块的方法模拟围岩松动圈的影响,各分区的岩体力学参数不同并作为待反演参数.此法是反分析中考虑围岩松动圈影响的一种新途径.通过某大型水电站工程地下厂房的实例进行验证,得到比较合理有效的结果,反演得到的物理力学参数可信,可用于后期预测.该研究为地下工程反分析提供了一些切实可行的方法和思路.

There are different characteristics between particle swarm optimization algorithm and variable metric optimization algorithm. A modified particle swarm optimization algorithm is proposed according to their compatibility and algorithm complementary. Then influence of loosing zone is analyzed through area division method based on loosing zone＇s formation mechanism, which is a new approach to consider the influence of loosing zone in underground engineering back analysis. The rock mass parameters of all areas are different and are treated as unknown parameters. These methods are applied to the underground power- house of a large-scale hydropower engineering. The computational results show that the proposed algo- rithm is reasonable and effective. The parameters calculated by back analysis are believable and can be used to forecast anaphase excavation. The results provide some feasible methods and considerations for under- ground engineering back analysis.