折叠网壳结构风压分布特性的神经网络预测研究

Prediction of wind pressure distribution on folding reticulated shell structure using neural networks

为了全面分析轻型可移动折叠网壳结构的风压分布特性,基于BP神经网络的基本原理,运用Matlab软件对野营折叠网壳结构风压分布特性进行了神经网络预测;并结合折叠网壳结构风洞试验,分析了其在不同风速和不同风向角下的风压系数分布规律和表面风压分布特性,将结果与风洞试验进行对比,两者吻合较好。结果表明:将风洞试验技术与神经网络方法相结合,基于有限的风洞试验数据可以预测结构未知点的风压系数;通过对测点进行插值计算可以全面掌握结构的风压分布特征,为结构抗风的精细化分析和设计提供一个有效方法。这为轻型可移动折叠网壳结构安全合理设计和改进建筑气动外形提供了理论依据。

To detailedly investigate the wind pressure distribution characteristics of light removable folding reticulated shell, a BP neural network-based prediction is conducted by using software Matlab for a folding camping shell structure. Meanwhile, distributing laws of the wind-pressure coefficients and the surface wind pressure under different wind speeds and wind directions are analyzed, which agree well with the wind tunnel test data. The results show that: through combining the wind tunnel test technique with the neural network method, the wind-pressure at any point on reticulated structures can be reasonably predicted according to the limited wind tunnel test data and the wind-pressure distributions of whole structure can be obtained based on the interpolation between measuring points. The method proposed in this paper provides an effective approach to fine analysis and wind-resistant design of structures, and it is applicable to the safety design and aerodynamic optimization of light removable folding reticulated shell.