高边坡开挖变形多目标全过程反分析方法

Multi-objective whole-process inverse analysis method for excavation deformation of high slope

边坡工程通过预测开挖后的位移变化来评估开挖风险,岩体工程参数直接关系到最终结果。采用边坡变形的多目标反分析方法获取准确合理的岩体工程参数,该方法基于非劣排序遗传算法(non-inferior sorting genetic algorithm Ⅲ,NSGA-Ⅲ),采用经过训练的BP神经网络(back propagation neural network,BPNN)代替数值计算,采用NSGA-Ⅲ用于搜索多目标方程的Pareto解集。以锦屏一级水电站左岸坝肩边坡为依托验证所提方法,考虑外观测点的水平位移及竖直位移、坡表浅部位移、深部裂缝变形共4种现场数据,选择对位移变化有较大敏感性的5类岩体材料参数作为反演参数,在不同目标函数组合下进行参数反演。结果表明:采用NSGA-Ⅲ的反分析方法能够综合利用多种监测数据,提高反演结果的可靠性和计算效率,利用反演获取的材料参数正向计算后所得变形与实际观测结果匹配良好,验证了该方法在边坡开挖工程中预测的准确性和可靠性。

Slope engineering assesses excavation risks by predicting post-excavation displacement changes,and the value of rock mass engineering parameters is directly related to the final results.To obtain accurate and reasonable rock mass engineering parameters,a multi-objective inverse analysis method of slope deformation is adopted.This method is based on non-inferior sorting genetic algorithm Ⅲ(NSGA-Ⅲ),employs trained back propagation neural network(BPNN)to replace numerical calculations,and utilize NSGA-Ⅲ to search for the Pareto solution sets of multi-objective equations.The proposed method is validated against the dam abutment slope on the left bank of the first stage of the Jinping Hydropower Station.Considering four types of on-site data,namely horizontal and vertical displacements of external monitoring points,shallow slope surface displacements,and deep crack deformations,five kinds of rock mass material parameters with significant sensitivity to displacement changes are selected as inversion parameters,and parameter inversion is conducted under different combinations of objective functions.The results indicate that the inverse analysis method of NSGA-Ⅲ can comprehensively utilize a variety of monitoring data,and improve the reliability and computational efficiency of the inversion results.After forward calculation using the obtained material parameters from the inversion,the deformation obtained matches well with actual observation results,which verifies the accuracy and reliability of the proposed deformation prediction method in slope excavation engineering.