结构损伤识别神经网络方法研究  

Study on Neural Network Method for Structural Damage Recognition

作　　者：尚钢[1] 饶文碧[2] 程洪斌[2] 

机构地区：[1]黄冈师范学院 [2]武汉理工大学

出　　处：《武汉理工大学学报》2002年第10期61-63,共3页Journal of Wuhan University of Technology

基　　金：湖北省教育厅 2 0 0 1重大科研项目资助 (2 0 0 1Z4 30 0 4)

摘　　要：阐述了利用神经网络建立结构损伤辨识系统的过程 ,并针对结构损伤识别中训练样本多、难以取舍的情况 ,采用了一种基于正交设计思想的样本选取方法 ,来构造网络训练所需的“最小”样本集。通过对缺陷板结构损伤识别的实例分析 ,说明了基于人工神经网络的损伤辨识系统的正确性。This paper presents general process of building the system of structure's damage identification based on the artificial neural network. In damage identification of structure, the training sample set is difficult to select, so this paper gives a solution by creating a 'minimum' sample set from the whole sample based on the method of cross experimental design. Using this method, a damage identification system of reinforced concrete board structure is set up, and the results of identification show that the system is available.

关 键 词：人工神经网络 结构损伤识别 样本选取 

分 类 号：O342[理学—固体力学] TP183[理学—力学]