大坝服役性态时空多尺度递阶评估模型和诊断方法

“无人值班、少人值守”现代大坝管理目标推动了大坝安全智能监控和预警的诉求。拟基于知识发现理念，借助坝工与人工智能等学科交叉，开展大坝安全全过程自适应监控、大坝材料特性衰减过程和大坝系统安全度演化过程静动联合反演、大坝服役性态转异成因综合诊断和预警等模型、方法研究，以实现人为参与度低的大坝宏微观性态时空多尺度递阶评估及诊断：(1)研究大坝安全单测点抗噪和多测点时空场监控模型，提出去噪与建模一体化实现方法和模型解析技术；(2)研究大坝材料特性参数静动联合反演方法，建立材料特性衰减过程模型，提出模型实时校正技术；(3)研究基于多参量的大坝系统安全度演化过程模型推理方法，提出大坝系统安全度转异判据；(4)研究大坝服役性态转异成因综合诊断方法，提出性态转异的点、线、面联合预警指标；(5)研究小波、支持向量机等知识发现方法的大坝监控领域适宜性集成机制和改进技术。发表论文20篇、专著1部、发明专利4项。

Based on an idea of knowledge discovery, artificial intelligence and dam engineering are combined to implement the multi-scale hierarchical evaluation and diagnosis of macro and micro behavior for in-service dam. The research interest is focused on the self-adaptive monitoring models of dam safety, the back analysis methods of time varying process on material parameters and system safety, the diagnosis methods and early-warning criterion of abnormal behavior. (1) The models monitoring safety of one point and space is studied. The method is proposed to implement simultaneously denoising and modeling. The way is developed to separate the model. (2) The method is studied to implement the static and dynamic data-base back analysis for material parameters. The model of time varying process on material parameter is built. The real time correction way of above model is proposed. (3) The method is studied to inference the time varying process model on dam system safety. (4) The diagnosis method of dam abnormal behavior is studied. The point-line-field early-warning criterion of abnormal behavior is developed. (5)Aiming at adaptability in the field of the early-warning technique, the integration mechanism and technical improvement of wavelet and SVM are established.