中文摘要：

考虑岩土体渗流场在工程施工及运行过程中的动态特征,依据水头和流量2类测点的实测时间序列数据,建立基于正交设计、非稳定渗流正分析、BP神经网络和遗传算法相结合的非稳定渗流场反演分析方法,在一定程度上体现了多目标优化、全过程反演的特点,较好地解决了岩体渗透系数反分析的唯一性与可靠性问题。为了避免在反问题中由于利用水头和流量2种类型观测资料所带来的量纲问题,将水头和流量时间序列数据误差项归一化后构造量纲一化的目标函数,并在目标函数中引入权系数w,从而避免寻优时某一类型观测信息占优。针对长河坝水电站基坑施工期涌水问题,根据施工期各渗压计、基坑渗漏量实测时间序列资料和上游河水变化过程曲线,采用上述方法反演了坝基岩土体的渗透系数。反演分析结果表明:渗压、流量测点的实测和计算时间序列数据吻合较好,反演给出的各岩土体渗透系数与现场钻孔压水试验得到的岩体渗透特性分级相符,从而论证了该反演分析方法的可行性和反演结果的可靠性。研究结果为进一步评价大坝正常蓄水后坝基防渗体系的有效性提供了依据。

英文摘要：

A new approach for back analysis the rock/soil permeability based on the transient flow was developed considering the transient nature of the seepage flow in rocks and soils during construction and operation of engineering works. The proposed method utilizes the measured transient data of both piezometric head and flow rate is based on the orthogonal design,the finite element analysis of transient seepage flow,the BP(back propagation) neural network and the genetic algorithm. In order to avoid the dimensional inconsistency induced by different types of measurements,normalized error terms of the transient pressure head and seepage flow rate are used for constructing the objective function,and a weight coefficientw is introduced to balance the contributions by the piezometric head and flow rate terms. The proposed method is applied to back calculate the hydraulic conductivity around the foundation pit for construction of the rockfill dam with gravel wall core on Changhe river. The measured water leakage is much higher than the one expected in the design period. The back-calculated piezometric head and flow rate are in good agreement with the measured results. The back-calculated hydraulic conductivities of the foundation rocks are strictly within the ranges of permeability of graded rock mass determined with the in site borehole packer tests,which demonstrates the effectiveness of the back-analysis method and the reliability of the back-calculated results.