基于人工神经网络的典型桥梁断面气动参数识别

基于同济大学风洞实验室既有大跨度桥梁试验数据成果，利用Access数据库软件和Java编程语言，集成了大跨度桥梁抗风性能的数据库系统。通过人工神经网络技术对人工神经元的训练和神经元间连接权值的调整，建立大跨度桥梁主梁气动参数（包括静力三分力系数和颤振导数）的智能化识别方法，主要针对扁平箱梁和倒梯形箱梁两种断面。气动参数的神经网络输出与期望输出间的误差符合预期要求，以期可作为桥梁结构初步设计阶段参考。

Aerodynamic parameter identification of typical bridge sections based on artificial neural network

Based on the existing experimental results of long-span bridges in the wind tunnel laboratory of Tongji University, a database system for wind-resistant performance of long-span bridges is integrated by Access Database software and Java programming language. Based on the artificial neural network technology, artificial neurons were trained and neuron connection weights were adjusted, and consequently the intelligent identification method of aerodynamic parameters (including aerostatic coefficients and aerodynamic derivatives) was established. This method is mainly aiming at flat box girder section and inverted trapezoidal box girder section. The errors between the neural network output and the expected output can meet the expected requirement, and the prediction results can be referred in the preliminary wind-resistant design of bridge structures.