Vibration-based methods can be used to detect damage in a structure as its vibration characteristics change with physical changes in the structure. Arch bridge is a popular type of bridge with rather complex vibration characteristics which pose a challenge for using existing vibration-based methods to detect damage in the bridge. Further, its particular geometry with a curved arch rib and vertical members (either in compression or tension) to support the horizontal deck makes the process of damage quantification using vibration-based methods harder and challenging. This paper develops and presents a vibration-based method that utilizes damage pattern changes in frequency response functions (FRFs) and artificial neural networks (ANNs) to locate and quantify damage in the rib of deck-type arch bridge, which is the most important load bearing component in the bridge. Principal component analysis, which is performed to reduce the dimension of original FRF data series and to obtain limited principal component analysis (PCA)-compressed FRF data is used in the development of the proposed method. FRF change, which is the difference in the FRF data between the intact and the damaged structure, is compressed to a few principal components and fed to ANNs to predict the location and severity of structural damage. The process and the hierarchy of developed ANN systems are presented, including the “fusion network” concept, which individually analyses FRF-based damage indicators separated by sensor locations. Finally, results obtained for many tested damage cases (inverse problems) are presented, which demonstrate the applicability of the proposed method for locating and quantifying damage in the rib of deck type arch bridge.

中文翻译：

---

使用频率响应函数和人工神经网络定位和量化甲板型拱桥的损坏

基于振动的方法可用于检测结构中的损坏，因为其振动特性会随着结构中的物理变化而变化。拱桥是一种流行的桥梁类型，具有相当复杂的振动特性，这对使用现有的基于振动的方法来检测桥梁中的损伤提出了挑战。此外，其特殊的几何形状具有弯曲的拱肋和垂直构件（无论是受压还是受拉）来支撑水平甲板，这使得使用基于振动的方法进行损伤量化的过程更加困难和具有挑战性。本文开发并提出了一种基于振动的方法，该方法利用频率响应函数 (FRF) 和人工神经网络 (ANN) 中的损伤模式变化来定位和量化桥面拱桥肋骨的损伤，它是桥梁中最重要的承重部件。主成分分析用于降低原始 FRF 数据序列的维数并获得有限主成分分析 (PCA) 压缩的 FRF 数据，用于开发所提出的方法。FRF 变化，即完整结构和受损结构之间 FRF 数据的差异，被压缩为几个主成分并馈送到 ANN 以预测结构损坏的位置和严重程度。介绍了开发的人工神经网络系统的过程和层次结构，包括“融合网络”概念，它单独分析由传感器位置分隔的基于频响函数的损伤指标。最后，给出了许多测试损坏情况（逆问题）获得的结果，