【作者】 徐爱敏；

【导师】 徐兴； 叶贵如；

【作者基本信息】 浙江大学 ， 桥梁与隧道工程， 2008， 博士

【摘要】 耐久性问题以及结构健康评估是跨海大桥安全运营中最重要内容之一。本文以杭州湾跨海大桥70米预应力混凝土箱梁结构的耐久性评估和健康监测为对象,根据结构特征和杭州湾的海域环境特点,应用混凝土结构基本理论、力学原理、系统理论、测试技术、信号分析处理、图形学、信息网络、耐久性、可靠性、动力损伤识别、人工神经网络以及概率统计分析等理论方法分析了结构使用寿命的主要影响因素以及耐久性失效概率随服役时间的变化规律,研究设计了跨海桥梁的健康监测系统,并对信号处理技术和损伤识别方法等进行了探讨,提出了采用HHT变换进行非平稳信号处理的理论基础和分析技术,建立了基于人工神经网络的70米箱梁结构动力损伤诊断系统,为大桥运营管理提供了理论依据和测试、信号分析方法。本文通过一系列研究工作,得出了以下几个方面的结论:(1)基于混凝土保护层厚度、氯离子初始浓度、氯离子扩散系数、氯离子临界浓度以及结构表面氯离子浓度等影响因素的概率分布特征,提出了70米箱梁结构耐久使用寿命的统计分布特征和概率密度函数。给出了70米箱梁结构耐久性失效概率随服役时间的变化规律,并基于不同抗力水平对70米箱梁结构的脱钝前锋面进行了随机动态可靠性预测分析,为70m箱梁结构的耐久性评估建立了理论预测手段。(2)提出了适用于大型跨海桥梁健康监控系统以及信号采集、传输技术。阐述了健康监控系统的总体概况、传感器子系统、信号的采集与传输、处理与控制子系统以及结构健康综合子系统的设计原则、内容及实施方法。(3)基于响应面方法建立了信号采样点间的递推关系,解决了HHT信号处理中的边界延拓问题;利用奇异值理论,证明了由最小二乘问题导出的线性方程组的基于奇异值分解得到的广义逆是原线性最小二乘问题的解,解决了目标函数的Hessian矩阵非正定时所存在的收敛性问题。改进了利用HHT诊断实际桥梁结构损伤的精度,为该技术在实际工程应用的可行性进行了有益的探索。(4)基于杭州湾跨海大桥70米箱梁结构的设计构造特点,建立了基于人工神经网络的70米箱梁结构动力损伤诊断系统,可有效识别杭州湾跨海大桥70米箱梁结构局部刚度损伤的位置和程度,补充和完善了杭州湾跨海大桥70米箱梁结构健康监控系统,为确保杭州湾跨海大桥的正常安全使用,具有一定的理论和工程实际应用意义。更多还原

【Abstract】 Durability and structural health evaluation are two important factors influencing the safety of the bay bridge. Hence the research focuses on the durability and health evaluation of the 70m box girder of the Hangzhou Bay Bridge. Considering the structural and environment characteristics of the Bridge, the study analyzes main influencing factors of the structural service life and the failure probability laws by using the concrete theory, the mechanics, the systematic theory, the testing technique, the signal processing, the information network, the durability theory, the dependability theory, the dynamic damage identification system, the artificial neural network and the probability statistical analysis. The study also designs the health monitoring system and discusses signal processing techniques and damage diagnosing methods. Further, the theory foundation and the processing technique of HHT signal processing method are put forward and the artificial neural network of the 70m box girder structural dynamic damage diagnosing system is founded. All these will provide theoretical foundations, measuring methods and signal processing methods for the service management of the Bridge.By all the studies, the following conclusions are drawn.Firstly, by studying on the thickness of concrete cover, the Cl~- initial density, the Cl~- diffusion coefficient, the Cl~- critical density, the Cl~- density on the structural surface and their probability distribution characters, the statistical distribution characteristics and probability density function are derived. Besides, the law of the durability failure probability of the 70m box girder is presented and then the random dynamic probability is forecasted which can be used as theoretical method of the durability evaluation of the 70m box girder.Secondly, the signal collection and transmission technique which is suitable for the health monitoring of the large span bay bridge are brought forward. At the same time, the general survey, sensor sub-system, signal collection and transmission, signal processing sub-system, design principle and application method of the structural health comprehensive sub-system are expatiated.Thirdly, HHT is used to analyze the nonlinear and non-stationary signals for diagnosing the faults in bridge structure. How to extend the boundaries of the analyzed signal for sifting process is a key problem of HHT. And a new technique based on response surface method is presented to deal with the difficult problem. It has been proved that the existing methods, that is AR model method and linear neutral net method, are special cases of the new generalized method. The boundary extension problem arising from HHT can be described by mathematical programming, and traditional gradient algorithms may diverge when the Hessian matrix of the object function is non-positive. It has been proved that the solution of the original programming problem can be obtained by solving linear equations which is solved by SVD. HHT is used to diagnose the damage of the practical bridge, and analysis results show that the method with new boundary extension technique performs successfully.At last, the dynamic damage diagnosing system of the 70m box girder is founded on the basis of the artificial neural network by considering the structural design and construction characteristics of the 70m box girder. This system can identify the position and extent of the local damage effectively and consequently complement the structural health monitoring system of the 70m box girder. Besides, it can guarantee the Hangzhou Bay Bridge’s safety during service and thereby has significant theoretical and practical meaning.更多还原