基于概率神经网络的斜拉桥子结构损伤识别研究

It is an active research subject in recent years that the artificial neural network has been used on structure damage identification. The probabilistic neural network is suitable to pattern classification. In this paper, a method of substructure damage identification and the theories of probabilistic neural network are introduced firstly. Then, the numerical simulation for the damage identification of a cable-stayed bridge is carried out using probabilistic neural networks, in which the training simples and test simples are under noise-free and including the noise. Finally, identification precision and noise effect of the neural networks are compared. According to the analysis above, the conclusion is given that the probabilistic neural network is suitable to substructural damage identification of cable-stayed bridges.