土石坝渗透特性反演及损伤识别的不确定性理论与方法研究

The anti-seepage body is the important guarantee of the dam safety operation, there are some cracks caused by many kinds of influencing factors such as the construction quality control, which is called the damage in this dissertation. The seepage properties will change caused by the damage, In the same time, there have uncertainty of the materials, loads, geometry size, boundary conditions in earth rockfill dam. Considering these uncertainty and difficulty, by using the cloud theory which have the advantage of handle fuzziness and randomness, and the advantage of transition between qualitatives and quantitatives, the coupling model will be established to explore the uncertainty in earth rockfill dam and to established the back analysis of seepage properties. On this basis, the damage of the anti-seepage body will be determined according to the variation of seepage properties. The research contents included: First, study on the spatial variability of permeability coefficient by cloud theory, and the cloud distribution of permeability coefficient can be established; second, the three-dimensional finite element model were established, used the seepage monitoring model, the seepage parameter can be back analysed by cloud theory; After that, Using the thought and method of back analysis for reference ,the model of damage identification of clay filled cut-off, core wall and curtain wall was established, and the health status of the key parts can be diagnosed. The research can provide theoretical basis for the health diagnosis and the optimization design of the earth rockfill dam, and can provide decision basis and technical support for safe operation and danger control of the earth rockfill dam, which have important theoretical value.

防渗体是土石坝安全运行的重要保证，由于施工质量控制等诸多因素的影响，防渗体内或多或少存在缺陷，称之为损伤。损伤必然会降低防渗体的功能，表现为渗透特性的改变。同时，鉴于理论计算与分析还存在着材料、荷载、几何尺寸和边界条件的不确定性等问题，本项目拟采用能够兼顾考虑模糊性、随机性及二者关联性特点而发展起来的定性定量转换模型的云理论，建立基于云发生器的耦合模型，探究土石坝渗透特性反演模型计算及监控数据中的随机性、模糊性等不确定性，并根据防渗体渗透特性的变化确定其损伤情况。研究内容包括：研究土石坝渗透特性的空间变异性，分析土石坝渗透系数的云分布特性；建立三维有限元模型，利用渗流监控资料，采用云理论方法反演关键部位的渗透参数；应用反分析的理论方法，建立土石坝防渗体的损伤识别模型。研究既可为土石坝的优化设计和健康诊断提供理论依据，又可为大坝的安全运行和除险加固提供决策依据和技术支撑，具有重要的理论价值。

我国是世界筑坝大国， 无论筑坝数量和规模均居世界第一， 在已建的 9.8万多座大坝中，绝大部分为土石坝，这些水利水电工程在防洪、灌溉或供水、发电和航运等方面产生了巨大的社会经济效益。然而，我国水库工程的安全状况并不乐观。我国一半以上水库建成于 20 世纪 50 年代至 70 年代，经过几十年的运行，大多已处于病险状态，其中一大部分是由于渗透破坏引起的，因此进行土石坝渗透特性的研究非常有必要。同时，鉴于理论计算与分析还存在着材料、荷载、几何尺寸和边界条件的不确定性等问题，也急需对土石坝渗透特性的分布情况进行研究。因此基于能够兼顾考虑模糊性、随机性及二者关联性特点而发展起来的定性定量转换模型的云理论，项目首先进行了土体渗透特性参数的土工试验，对试验数据进行检验以研究渗透参数的云分布情况，确定了渗透参数服从泛正态云分布；其次，建立了防渗体的渗流有限元模型，构建正向云发生器、云推理算法，并利用监控资料进行了防渗体渗透特性的反演分析；最后，建立了基于云理论和神经网络法相结合的云神经网络耦合模型，研究土石坝防渗体的损伤表现与实际监测资料之间的耦合关系，并评价了特定水库大坝在其服役期内的健康状况。项目的研究成果既为土石坝的优化设计和健康诊断提供了理论依据，又为大坝的安全运行和除险加固提供了决策依据和技术支撑，具有重要的理论价值。