径向基函数神经网络在桥梁结构损伤识别中的应用  

Application of RBF Neural Network in Bridge Structure Damage Identification

作　　者：路淑芳 谭祥 刘旺 LU Shufang;TAN Xiang;LIU Wang(CSCEC 7th Division International Engineering Co.,Ltd.,Guangzhou Guangdong 510400,China)

机构地区：[1]中建七局国际工程建设有限公司,广东广州510400

出　　处：《建筑施工》2022年第2期394-399,共6页Building Construction

摘　　要：鉴于新旧桥梁在日常运营中缺乏连续、实时的损伤情况分析,基于桥梁结构损伤识别与机器学习中的径向基函数理论,提出了2种损伤识别方法。一是分步识别法,运用频率的变式识别损伤位置,曲率、频率组合输入识别损伤程度,得到了88%以上的位置识别正确率与92%以上的程度识别正确率,适用于对准确率要求较高的结构损伤识别。二是综合识别法,在分步识别法的基础上对网络结构进行优化,利用神经网络的输出特征可直接判别损伤位置和损伤程度,得到了78%以上的识别准确率,适用于桥梁结构大数据的在线分析。试验证明了2种径向基函数(radial basis function,RBF)神经网络可以用于桥梁不同实际条件下的损伤情况识别,对于神经网络在桥梁结构损伤识别方面有更加准确的认识和分析,为以后进一步研究神经网络在桥梁结构损伤识别方面提供参考。In view of the lack of continuous and real-time damage analysis of old and new bridges in daily operation,two damage identification methods are proposed based on the radial basis function theory in bridge structural damage identification and machine learning.One is the step-by-step identification method,which uses the variation of frequency to identify the damage location and the combination of curvature and frequency to identify the damage degree,and obtains more than 88% of the location identification accuracy and more than 92% of the degree identification accuracy,which is suitable for structural damage identification with high accuracy requirements.Another is the comprehensive identification method,which optimizes the network structure on the basis of the step-by-step identification method.Using the output characteristics of the neural network,the damage location and damage degree can be directly identified,and the identification accuracy of more than 78% is obtained,which is suitable for the online analysis of big data of bridge structure.The test shows that the two radial basis function(RBF)neural networks can be used to identify the damage of bridges under different actual conditions.It has a more accurate understanding and analysis of neural network in bridge structure damage identification,which provides a reference for further research on neural network in bridge structure damage identification in the future.

关 键 词：桥梁工程 损伤识别 径向基函数神经网络 分步识别法 综合识别法 

分 类 号：TU99[建筑科学—市政工程]