基于自由响应信号与BP神经网络的结构损伤程度识别研究

Study on structural damage degree identification based on free response signal and BP neural network

提出了基于自由响应信号与BP神经网络的结构损伤程度识别方法:首先采用随机减量技术提取自由响应信号,利用AR模型进行动态建模,计算结构损伤前后的模型系数之间的欧式距离;其后以结构损伤前后的欧式距离作为特征向量,输入BP神经网络确定结构的损伤程度。四层海洋平台结构的数值模拟表明,该方法是可行的,且具有一定的抗噪声能力。

The method for structural damage degree identification based on free response signal and BP neural network is put forward： Firstly, the random decrement technique is used to obtain the free response signal and the AR model is adopted to create the dynamic model and calculate the Euclidean distance between model coefficients before and after structural damage; Secondly, the Euclidean distance before and after structural damage is used as the characteristic vector, which is put into BP neural network to determine the degree of structural damage. The numerical simulation of a four-floor offshore platform shows that the proposed method is feasible and has a certain noise resisting ability.