基于环境激励的泄流结构多级信息融合损伤诊断方法研究

By the reasons of loads, environmental changes, material aging and other factors in the long running process, varying degrees of damage is produced in flood discharge structure inevitably. How to effectively detect the damage location and degree of flood discharge structure and timely warning are crucial to ensure the overall security of hydraulic structure. In order to solve the problems in the study of non-destructive testing of flood discharge structure, especially in the problem of underwater position detection, the project is planned to study on multilevel information fusion damage diagnosis of flood discharge structure base on ambient vibration by model test, prototype observation and finite element method. By hydroelastic model test, characteristic of dynamical response of flood discharge structures under ambient excitation is elucidated, the noise reduction filtering method and modal parameter identification method of strong background noise strong interference test signal are proposed, and the damage law and damage mode of flood discharge structure are revealed. According to the model test and prototype test data, combined with the damage law and damage mode, the damage sensitive characteristic factors are selected by multilevel information fusion and intelligent optimization algorithm, and the nonlinear relationship between damage sensitive characteristic factors and damaged mode is revealed, and then the methods of flood discharge structure on damage location and damage degree are achieved. These works can provide a basis for a comprehensive, accurately grasp of the flood discharge structure working state or damage change trend.

泄流结构在长期运行过程中受荷载、环境因素等作用及结构老化等原因，不可避免会存在不同程度的损伤，如何有效地检测出泄流结构的损伤位置及损伤程度对确保水工结构整体安全至关重要。针对泄流结构损伤检测中存在的不足，尤其水下位置检测难的问题，本项目拟通过水弹性模型试验、原型观测和有限元数值分析相结合的方法，开展基于环境激励的泄流结构多级信息融合损伤诊断方法的研究。通过水弹性模型试验，阐明环境激励下的泄流结构动力响应特性，提出强背景噪声强干扰测试信号的降噪滤波方法和模态参数识别方法，揭示泄流结构的损伤规律和损伤模式；根据模型试验、原型试验数据，结合损伤规律和损伤模式信息，采用多级信息融合技术和智能优化算法，甄选损伤敏感特性量，揭示损伤敏感特性量与损伤模式之间的非线性关系，实现泄流结构损伤位置和程度的识别。研究成果可为全面、准确地掌握泄流结构的工作性态或损伤变化趋势提供依据。

泄流结构在长期运行过程中受荷载、环境因素等作用及结构老化等原因，不可避免会存在不同程度的损伤，如何有效地检测出泄流结构的损伤位置及损伤程度对确保水工结构整体安全至关重要。本项目通过水弹性模型试验、原型观测和有限元数值分析相结合的方法，开展基于环境激励的泄流结构多级信息融合损伤诊断方法的研究。（1）研究并提出了一种基于经验小波变换的改进去噪方法，并用模拟信号验证了该算法具有很好的去噪效果。提出了基于NExT-ERA的奇异熵联合经验小波去噪的模态识别方法，能有效避免传统方法在系统脉冲响应获取、识别精度易受噪声干扰以及系统定阶等方面的缺陷。（2）提出相关性方差贡献率的数据层信息融合方法，将该方法应用到原型观测和模型试验中，均验证了该方法的有效性；（3）最后基于信息融合技术，依据D-S证据理论，提出了改进的基于Pignistic概率距离与冲突系数的D-S证据融合方法，该方法具有较好的收敛性和识别精度。