桥梁结构的损伤检测与识别技术研究56349摘 要 随着桥梁建设的持续发展，桥梁结构的形式和功能也日趋复杂，桥梁的修补和加固也越来越受到关注。 桥梁建成通车后， 由于受气候、 环境因素以及人为因素的影响，结构材料会被腐蚀和逐渐老化，长期的静、动力荷载作用，使其强度和刚度随着时间的增加而降低。 这不仅会更会使桥梁的使用寿命缩短， 更严重的会影响交通行车安全，危机人的生命。桥梁结构的检测、监测作为结构安全养护、正常使用的保证措施之一受到关注，如何对桥梁结构进行质量检测和安全监测也已成为国内外学术界、工程界研究的热点。 本课题主要针对目前在役桥梁存在的种种安全隐患，展开桥梁性能检测和损伤识别技术的研究，以随时了解桥梁的结构性能和安全情况，避免灾难性事故的发生。主要研究工作包括以下几个方面： 1.研究了遗传算法优化的神经网络在桥梁结构损伤检测中的应用。基于遗传算法优化的一种有效的结构损伤检测，建立评估钢筋锈蚀程度的人工神经网络模型。选择对结构损伤较为敏感的参数作为网络的输入向量， 结构的损伤状态作为网络的输出向量，建立损伤训练样本集，利用遗传算法优化神经网络的权值和阈值，进而预测结构损伤程度。实验发现通过此方法能较好的预测结构的内部损伤，一定程度上解决了对钢筋混凝土锈蚀损伤程度的评估。 2. 研究了基于挠度信息的桥梁裂纹局部定位方法。裂纹的出现会引起桥梁结构某些特征参数的不规则变化或突变,这些参数中包含着重要的损伤信息。结构裂纹的出现会引起结构中的局部应力集中， 导致空间域上结构的挠度信息在这一区域发生突变。因此，我们提取含裂纹的悬臂梁在静荷载作用下各节点的挠度值，作为神经网络的输入向量，以裂纹点的相对位置作为神经网络的输出，建立了神经网络模型，达到了很好的分类效果，对于损伤的快速局部定位取得了很好的效果。 3. 利用多表达式编程（Multi-Expression Programming, MEP）和频率等高线相结合的方法对桥梁结构裂纹定位进行了分析和研究。 将结构固有频率作为裂纹的诊断参数，通过求解裂纹有限元模型，对多表达式编程模型进行训练，并拟合出以裂纹相对位置和相对深度为自变量，裂纹结构固有频率为因变量的解曲面，将实测结构的前三阶固有频率作为输入，绘制出各阶频率等高线，根据频率等高线的交点及 MEP 的预测结果来定位结构的裂纹。 实验结果表明该方法能够有效的融合二者的优点实现裂纹 快速、准确定位。 关键词：桥梁检测，神经网络，多表达式编程，频率等高线 ABSTRACT With the continuous development of bridge construction, functions and formats of bridge become more complicated day by day. The technology of repairing and strengthening of bridge has been emphasized. After bridges having been constructed and opened to traffic, their material will be deteriorated or aged gradually because the influence of the weather, environmental factors and their strength and stiffness will degrade with the time running for action of the static and active loads applying on them. Not only will this endanger the safety of the traffic, but also it will shorten the life span of the bridge and claimed the lives of people. Therefore, detection and inspection of bridges have become one important guarantee that ensures safety maintenance and normal use of it. So, how to perform quality detection and safety inspection with bridges has become a research hotshot of foreign academia and engineering. View of the various hidden trouble of the in-service bridges exist, this paper presents some methods on crack fault diagnosis. We should learn the performance and the security situation of the bridge at any time, and avoid the catastrophic accidents happened. The main work and achievement are summarized as follows: 1. Application of genetic algorithms and neural networks in detection technique.We establish the artificial neural networks model to appraise the steel bar corrosion degree and chose some parameters which sensitive to the structure damage as the input vector of the network. This method takes the damaged structure condition as the output vector of the network, and uses the Genetic Algorithm optimization the network weight and threshold value. The main idea is establishing a damage training sample sets, and carries on the training to the network, then forecast structure extent of damage. 2. Local positioning of crack fault based on deflection information. Owing to the emergence of crack, some of the characteristic parameters of bridges will cause the irregular change or mutation, these parameters contains important information of the crack. Structural cracks can cause the appearance of the local stress concentration and the mutation of the deflection information in the same region of space domain. Therefore, we set up a neural network model which chooses the deflection information of cantilever beam under the static load and the relative positions of the crack as the input and output vector respectively. It is showed that the proposed method is feasible to diagnose the crack fault of structure. 3. A method of crack fault diagnosis based MEP (multi-expression programming) and frequencies contour line is discussed. In this method, the MEP model is trained by the cracks diagnostic parameters which are the inherent frequencies obtained by ANSYS. Based on the independent variables, the relative position and the relative depth of crack, the frequency surface is drawn. By taking the first three inherent frequencies of structure as input parameters (of the frequency surface), Contour Lines of frequencies is drawn. According to the intersection point of frequency contour lines and the forecasted results of MEP, the structural crack position is located. The results show that the method can be an effective integration of both advantages and achieve the rapid and accurate location of the crack. Keywords： Bridge Detection， Artificial Neural Network， Multi Expression Programming，Frequencies Contour Lines. 1 第一章 绪论 1.1 课题的研究背景和意义 桥梁在长期的使用过程中不免会发生各种结构损伤。损伤的原因可能是使用、维护不当、车祸事故等人为因素,也可能是地震、风暴等自然灾害。此外某些要道上交通量以大大高于预测流量的速度猛增也加剧了桥梁结构的自然老化。 这些因素均导致了桥梁承载能力和耐久性的降低,甚至影响到运营的安全。由此而引起的一系列问题都需要相应的维修、改造和加固来解决。而这些工作又必须在对桥梁结构详细和系统的检测的基础上才能妥善的进行1。 近年来，国内外桥梁倒塌事故逐年增加。1967 年 12 月，俄亥俄河上的一座主要桥梁倒塌，