基于神经网络的桥梁损伤位置识别

损伤位置识别是对大型桥梁结构进行损伤检测的重要一步.以汲水门斜拉桥为背景,对应用神经网络的模式分类技术识别桥面结构损伤位置的方法进行了研究.使用了两种类型的网络,简称动态网络和GA网络,探讨了这一方法的可行性.动态网络的网络结构是根据训练进程而动态地确定的.GA网络是在训练中引进了遗传算法.比较了两种网络对损伤位置的识别效果.结果表明,应用神经网络的模式分类技术对桥梁桥面结构损伤位置识别的方法是可行的.只需要较少的输入数据,两种网络均可产生较好的识别结果.

IDENTIFICATION OF DAMAGE LOCATION IN BRIDGE DECK BY NEURAL NETWORK

Identification of damage locations is an important step in damage detection for large-scale bridge structures. By taking the cable-stayed Kap Shui Mun Bridge as an example, a method of damage localization for bridge deck by pattern recognition technique of neural network is studied. Two kinds of networks, dynamic network and GA network, are used to investigate the feasibility of the method. For dynamic network, the network structure is constructed dynamically with training progress. For GA network, the genetic algorithm is introduced in network training. The recognition results by using the two networks are compared. The results show the feasibility of the proposed methods. Both networks can give a satisfactory result when a small number of parameters are inputted.