关键词: 地下厂房, 最小二乘支持向量机, 粒子群算法, 有限元模拟, 位移增量, 反分析法

Abstract:

In order to accurately determine the surrounding rock parameters of underground powerhouse, i.e. deformation modulus, Poisson′s ratio, cohesion, angle of internal friction and lateral pressure coefficient, the back analysis method for surrounding rock parameters based on displacement increment is established by modern mathematical methods like orthogonal design, least squares support vector machine and particle swarm algorithm. According to engineering geological condition analysis, the crosssection of unit No.8 of CocaCodo Sinclair (CCS) Hydropower Station is selected as the study object, and the geological structural analysis model is established by 2D elasticplastic finite element method. The back analysis of mechanical characteristics of surrounding rock and geostress field of CCS Hydropower Station is carried out based on monitored displacement increments of layered excavation of underground powerhouse caverns. The study results show that the computation values of displacement increments caused by the excavation at layer Ⅵ and Ⅰ of main powerhouse and layer 4 and 1 of converter room coincide with that of the monitoring values, and the maximum relative error is less than 10%, which indicates that the back analysis method of least squares support vector machine and particle swarm algorithm is feasible and effective.

Key words: underground powerhouse, least squares support vector machine, particle swarm algorithm, finite element simulation, displacement increment, back analysis method