【作者】 张乾飞；

【导师】 吴中如；

【作者基本信息】 河海大学 ， 水工结构工程， 2002， 博士

【摘要】 本文针对与工程安全密切相关的渗流控制问题，利用分形理论、神经网络、遗传算法等前沿数学方法和渗流力学等力学方法，结合大坝原型观测资料，研究建立各种渗流监控模型的新理论、新方法，由此分析复杂渗流场的演变规律，并用正反分析相结合的方法对复杂渗流场转异特征进行了研究。其主要内容如下： (1)复杂渗流场的渗流特性及分析方法。首先，在裂隙岩体渗流基本特性研究的基础上，对裂隙岩体渗流的地质力学模型进行深入研究；然后，根据分形几何理论研究了渗透结构面的分形渗流特性；最后，提出了求解复杂渗流场自由面的改进截止负压法，并将其用于三维复杂非稳定渗流场的有限元计算，编制了基于分区混合介质模型的数值计算程序。 (2)复杂渗流场特殊监控模型研究。引入库水位及降雨的滞后影响函数，建立了等效水位和等效降雨的概念，其中库水位的滞后模式采用正态分布曲线，而降雨滞后模式则采用对数正态分布曲线，较好地模拟了库水位及降雨对大坝渗流场的滞后影响作用，由此建立了基于滞后效应的渗流监控模型；引入库水位等效变化速率的概念，通过三维渗流有限元计算建立了考虑库水位变化速率的渗流混合模型；最后对复杂渗流场基流效应进行分析，指出基流效应主要来源于水压分量和时效分量，并推求了大坝基流下包线。 (3)复杂渗流场演变规律研究。对复杂渗流场影响因素及其演变过程进行分析，论证了渗流要素随渗透系数和库水位变化的关系表达式，建立了能够隐式反映复杂渗流场演变规律的渗透系数时序模型，并探讨了利用大坝渗流场的基流效应来建立时序模型的具体实现方法和步骤。 (4)复杂渗流场转异特征研究。首先，对帷幕老化及排水失效机理进行了研究；然后，通过有限元数值分析方法，研究了复杂渗流场中防渗帷幕以及排水孔的转异特征，并推导了渗流要素随帷幕渗透系数变化的经验公式；利用时效和监控指标、监控模型等判别方法对复杂渗流场进行转异识别；最后应用回归分析法和神经网络法对转异时效进行计算。更多还原

【Abstract】 Aiming at seepage flow governing problems related to safety of engineering closely, some new models and methods are presented for seepage flow safety monitoring based on the advanced theories and methods of mathematics and mechanics, such as fractal theory, artificial neural network, genetic algorithm, seepage flow mechanics, and so on. Then, combing with prototype observation data, the evolvement rule and variation feature of complicated seepage flow field are analyzed by analysis and back analysis methods. The main contents are as follows:(1) The behaviors and analysis methods of complicated seepage flow field. Firstly, based on the research of basic behaviors of seepage flow field in fractured rock mass, the geo-mechanical models of seepage flow are studied deeply. Then, by way of fractal geometry theory, the geometric characteristics and hydraulic behaviors are studied of water-conduction fractures. And then, an improved cut-off negative pressure method is presented to determine the free water surface of complicated seepage flow field, and the method is applied to 3-D FEM analysis of unsteady seepage flow field, and relevant numerical simulation program is developed based on divisional hybrid medium model.(2) Research on the special monitoring models of complicated seepage flow field. The hysteresis influence functions of water level and rainfall are adopted to reflect their hysteresis effect on the seepage flow, and the equivalent water level and the equivalent rainfall are obtained separately through normal distribution curve and logarithmic normal distribution curve. And following this method, the seepage monitoring models based on hysteresis effect are established. The equivalent heaving speed is put forward for reflecting the influence of heaving speed of water level on the seepage flow field, and the seepage hybrid models are obtained by considering the heaving speed of water level through 3-D steady-unsteady seepage flow FEM analysis. In the end, the basic effect of complicated seepage flow field is analyzed, and the conclusion is obtained of seepage base flow mainly derived from water pressure component and time-dependent component, and the lower boundary lines are gained of seepage base flow in dams.(3) Research on the evolvement rule of complicated seepage flow field. The influence factors and evolvement process are analyzed of complicated seepage flow field. The relationship is demonstrated of seepage flow element varying with permeability coefficients and water level. The time series model of permeability coefficients, which can implicitly describe the evolvement rule of complicated seepage flow field, is established. The concrete realization approaches and steps are elaborated to construct time series model by taking advantage of seepage base flow.(4) Research on the variation feature of complicated seepage flow field. Firstly, the aging mechanisms of impervious curtain and the malfunction mechanisms of drainage are studied. Secondly, by means of seepage flow FEM computation, the variation features are analyzed of impervious curtain and drain hole in complicated seepage flow field. Accordingly, the experiential formulas are derived of seepage flow element varying with impervious curtain’s permeability coefficients. Thirdly, the variation feature is distinguished by the way of time-dependent component method, monitoring indices method and monitoring model method. In the end, the variational time-dependent component is obtained through regression analysis method and artificial neural network model.更多还原