碾压混凝土坝跨尺度力学行为演化机制及跟踪监控方法

According to the operation property evolution mechanism and service performance prediction of roller compacted concrete dam driven by multiple resultant forces, previous studies by applicant show that compacted layer degradation is an important inducement to aggravate the performance evolution of dam structures and damage evolution analysis is the critical step to interpret and monitor the dam performance degeneration. However, there are relatively few systemic studies about the evolution mechanism of cross-scale mechanical behavior and tracking monitoring method for roller compacted concrete dam at present. In this project, roller compacted concrete dam is taken as the object, comprehensive research methods, e.g., theoretical analysis, numerical simulation, model test and prototype observation, are adopted to study the evolution characteristics and driving mechanism of macroscopic and mesoscopic damages of roller compacted concrete dam under multi-force coupling and to reveal the inherent relationship between material performance evolution and its dependent factors. Furthermore, by analyzing the behavior characteristics of performance evolution in the existing damaged dams, cross-scale mechanical behavior simulation analysis model considering time-varying damage of roller compacted concrete dam would be constructed to reveal cross-scale evolution mechanism of time-varying damage in operation period. Based on the related attributes of dam monitoring effect quantities and influence quantities, multi-dimensional space-time monitoring model of typical performance indexes of roller compacted concrete dam would be established and multi-source, and multi-scale tracking monitoring method for the operation effectiveness evolution of roller compcated concrete dam would be studied and proposed, which provides theoretical basis and decision support for long-term service and operation management of roller compacted concrete dam.

针对多元合力驱动下碾压混凝土坝运行性能演化机理及服役效力预测，申请人前期研究表明，层面劣化是加剧坝身结构性能演变的重要诱因，其损伤演化分析又是诠释大坝性能退变并施以监控的关键步骤，目前有关碾压混凝土坝跨尺度力学行为演化机制及跟踪监控的系统研究相对较少。本项目以碾压混凝土坝为对象，采用理论分析、数值仿真、模型试验与原型观测相结合的研究方法，研究多力耦合作用下碾压混凝土宏细观损伤演变特征及驱动机制，揭示大坝材料性能演化与其依存诸因素的内在关联；更进一步，通过分析既有损伤大坝性能演变行为特征，建立考虑时变损伤的碾压混凝土坝跨尺度力学行为仿真分析模型，以期揭示运行期坝体时变损伤跨尺度演化机制；基于大坝监测效应量与影响量的关联属性，构建碾压混凝土坝典型性能指标多维时空监控模型，研究并提出碾压混凝土坝运行效能演化状态的多源多尺度跟踪监控方法，以此为碾压混凝土坝长效服役和运行管理提供理论依据与决策支持。

针对多元合力驱动下碾压混凝土坝运行性能演化机理及服役效力预测，本项目组融合经验小波变换理论与局部异常因子识别机制，提出了能合理识别并校正原型监测数据中异常值的方法；结合奇异值分解理论与数据回归方法，构建了监测数据缺值优化插补模型。为了减少不确定性对碾压混凝土坝实际工作行为评价的影响，需要进行基于监测数据的反演分析。本项目组碾压混凝土拱坝变形行为统计分析，结合并层等效理论，提出了基于麻雀算法优化BP神经网络模型的筑坝材料参数反分析模型，并编制了相应的实现程序。基于动位移监测与室内物理试验成果，综合采用有限元分析、支持向量机、响应面模型、改进的粒子群算法等方法，分别提出了软基水闸弹性模量与密度、拱坝及其坝基分区动弹性模量优化反演模型。就混凝土坝变形预测方面，综合应用数据挖掘、机器学习等方法，从因果分析模型建模残差角度，建立了考虑模型残差内蕴有效信息的混凝土拱坝单测点变形预测统计和混合模型；同时，针对拱坝变形的时空关联性，结合有限元分析与支持向量机算法，提出了考虑残差序列混沌效应的混凝土拱坝变形时空混合预测模型及其实施方法。最后，考虑到概率可靠性分析方法受不确定参数严格随机性要求、计算结果对参数过敏感性高及其功能函数高度非线性等因素制约，而实际工程中筑坝材料参数难以精确获取的问题，构建了考虑不确定参数界限的重力坝单元非概率可靠性指标计算模型，并融合Kriging模型提出了一种重力坝非概率可靠性分析方法；结合区间数学和非概率可靠性等理论和方法，构建了基于区间参数的重力坝单元与体系非概率可靠性计算模型，并提出了一种基于响应面方法的重力坝非概率可靠性指标计算方法。