论文部分内容阅读
渗流场参数的获取是研究运行期高心墙堆石坝渗流特性的难点之一。针对糯扎渡高心墙堆石坝,利用饱和–非饱和渗流场有限元程序生成学习样本,借助支持向量机的高度非线性映射能力,建立了渗透系数与水头之间的映射关系。再以识别误差目标函数为适应值,采用粒子群优化算法反馈搜索以建立大坝渗透系数反演模型。以大坝最大横剖面典型渗压计测点为实测点,采用一维固结理论推导了大坝心墙超静孔隙水压力消散计算公式,并对心墙水头实测值进行修正。通过对运行期库水位稳定时段渗流场的反演得到大坝待反演分区的渗透系数,再利用水位上升期对应的渗流场进行验证。结果表明,渗透系数反演结果是合理的。
The acquisition of seepage field parameters is one of the difficulties in studying the seepage characteristics of high core rockfill dam during operation. Aiming at Nuozhadu high core rockfill dam, the learning samples are generated by the finite element program of saturated - unsaturated seepage field and the mapping relationship between the permeability coefficient and the water head is established by means of the highly nonlinear mapping ability of support vector machines. Then, the objective function of identification error is used as the fitness value, and the particle swarm optimization algorithm is used to feed back the search to establish the dam inversion model. Taking the typical piezometer measuring point of the largest cross section of the dam as a measuring point, the formula of dissipating excess pore water pressure on the core wall of the dam is deduced by using the one-dimensional consolidation theory, and the measured value of the head of the core wall is corrected. Through the inversion of the seepage field during the stable reservoir water level period, the permeability coefficient of the area to be inverted is obtained, and the seepage field corresponding to the rising period of the water level is validated. The results show that the inversion of permeability coefficient is reasonable.