水力发电学报2017，Vol.36Issue(12)：78-86,9.DOI:10.11660/slfdxb.20171209
考虑实时施工质量因素的高心墙堆石坝沉降变形分析
Analysis of settlement deformation in high core rock-fill dams considering real-time construction quality factors
摘要
Abstract
Previous studies usually divide a statistical analysis model of the settlement in high core rockfill dams into filling factors,aging factors,and water pressure factors,taking a dam block as a homogeneous body.Such a simplification ignores the influence of nonuniformity in settlement deformation caused by the differences in construction quality.This study develops a settlement deformation analysis model for the dams of this type to take into account the real-time construction quality factors together with the factors of filling,aging and water pressure,based on an analysis of stress and deformation in the filling body.This multi-parameter nonlinear model is combined with a M5'model tree in simulation of the dynamics of settlement deformation.It combines the advantage of such a tree in solving global nonlinear problems and the fast convergence of a particle swarm optimization algorithm in solving multi-parameter problems,and thus it is able to fit and predict the settlement process.We compare it with a traditional analysis model excluding construction quality factors and a multiple linear regression analysis model.Results show that this new model is the best in terms of absolute errors and square sum errors in calculation of the settlement deformation and comprehensive index but its correlation degree remains as low as those of the other two models.This study suggests that modeling the real-time construction quality factors provides more reliable and objective analysis on the settlement deformation of high core rock-fill dams.关键词
高心墙堆石坝/沉降变形/施工质量/M5'模型树/粒子群算法Key words
high core rock-fill dam/settlement deformation/construction quality/M5'model tree/particle swarm optimization分类
建筑与水利引用本文复制引用
王飞,任炳昱,佟大威,鄢玉玲,王佳俊,钟登华..考虑实时施工质量因素的高心墙堆石坝沉降变形分析[J].水力发电学报,2017,36(12):78-86,9.基金项目
国家自然科学基金(51439005 （）
51621092 （）
51339003) （）
