基于RBF神经网络设计的桥梁结构损伤识别方法研究(英文)

Study on the Method of Bridge Structural Damage Identification Based on RBF Neural Network

首先根据桥梁结构的动力特性分析,构造了用于结构损伤识别的损伤标示量,并从理论上分析了该参数用于结构损伤识别的可行性.然后,从径向基函数(RBF)神经网络结构、网络设计和网络训练算法等方面论述了RBF神经网络理论,着重说明RBF网络的调用及径向基函数中心和宽度的确定步骤.最后,以一座装配式预应力钢筋混凝土系杆拱桥为工程实例,通过改变构件的弹性模量降低单元刚度来模拟结构损伤程度,并以任意三组向量对网络进行测试,说明了基于频率参数和RBF网络方法的结构损伤识别的可行性和准确性.

In this paper, the dynamic characteristics of bridge structure are analyzed firstly to obtain damage identification parameters, and the feasibility of these parameters to identify structural damage is investigated in theory. And then,the structure design and training algorithm of RBF neural network theory are discussed, emphsized in the calling function and the calculation steps of RBF center and width. A assembled PC tied arch bridge is simulated with numerical method by changing its Young＇s modulus to reduce element stiffness, and three vector groups are selected at random to test the RBF ANN. The feasibility and accuracy of damage identification of tied arch structure based on frequency parameters and RBF ANN are presented.