摘要
为实现非均布荷载工况或支撑不规则曲面屋盖情况下的树状结构找形设计,提出了一种基于BP神经网络算法的智能找形方法。该方法将树状结构各支撑点的荷载、跨度及分叉角度作为设计条件,以节点零弯矩和结构稳定性为设计目标,利用人工智能BP神经网络对下层分级节点进行智能定位,由上而下逐级递推实现树状结构的整体几何形态找形。以二维树状结构为例,阐述了智能找形的具体实现方法和步骤,并与逆吊递推找形的结果进行了对比分析。研究结果表明,智能找形方法不仅可以实现所有节点的零弯矩要求,而且分别降低了48.46%和29.01%的等效应力,具有可行性。此外,找形思路和BP神经网络的结合解决了树状结构在承受非均布荷载作用或支撑不规则曲面结构时找形定位难的问题,不仅增强了其应用方面的普适性,而且提升了整个找形过程的高效性和智能性。
In order to realize the intelligent shape finding design of tree-like structures under the conditions of bearing non-uniform load or supporting irregular curved roof,an intelligent shape finding method based on BP neural network is proposed.In this method,the design conditions are the load,span and bifurcation angle of each support point.The objective of shape finding design is to keep a zero bending moment and meet the structural stability requirements.This method can realize the intelligent positioning of hierarchical nodes by following the principle of recursive from top to bottom and by using artificial intelligence BP neural network.Taking the two-dimensional tree-like structures as examples,the realization method and steps of intelligent shape finding are described,and a comparative analysis with the results of reverse-hang recursive method is conducted.The results show that the intelligent shape finding method not only realizes the zero bending moment requirement of all joints,but also reduces the equivalent stress at joints by 48.46%and 29.01%respectively,which verifies the feasibility of this method.In addition,the combination of shape finding idea and BP neural network solves the shape finding problems that are difficult to complete the positioning of tree-like structures when they are subjected to uneven distribution load or support irregular curved surface structure,which not only enhances the universality in application,but also improves the efficiency and intelligence in the whole shape finding process.
作者
赵艳男
杜文风
王英奇
王辉
高博青
董石麟
ZHAO Yannan;DU Wenfeng;WANG Yingqi;WANG Hui;GAO Boqing;DONG Shilin(Institute of Steel and Space Structures,Henan University,Kaifeng 475004,China;Department of Civil Engineering,Zhejiang University,Hangzhou 310058,China)
出处
《建筑结构学报》
EI
CAS
CSCD
北大核心
2022年第4期77-85,共9页
Journal of Building Structures
基金
国家自然科学基金项目(U1704141,52178172)。
关键词
树状结构
神经网络
智能找形
静力分析
tree-like structure
neural network
intelligent shape finding
static analysis
