基于多源数据的结构安全性能智能化评估  

Intelligent evaluation of structural safety performance based on multi-source data

作　　者：朱昊梁[1] 王幼松[1] ZHU Haoliang;WANG Yousong(School of Civil Engineering and Transportation,South China University of Technology,Guangzhou 510640,China)

机构地区：[1]华南理工大学土木与交通学院,广州510640

出　　处：《建筑结构》2024年第23期27-34,共8页Building Structure

摘　　要：预应力钢结构构造复杂,安全性能涉及各类影响因素、设计参数及力学参数,为实现预应力钢结构安全状态的智能化评估,提出了基于多源数据的结构安全性能智能化评估方法。在研究过程中,将评估分为结构安全性能的分析与预测两个方面。首先建立多源数据与结构安全性能之间的关系,分别利用D-S证据理论与BP神经网络对结构安全进行分析与预测。在此基础上,融合数字孪生搭建了结构安全智能化评估模型并形成了实施流程。以弦支梁屋盖(对称结构)为试验验证对象,分析了温度作用下结构安全性能的变化,并预测了不同设计参数在荷载作用下对应的安全性能。通过安全评估方法,获取了结构的关键受力构件及设计参数与力学参数的关联机制,为结构安全维护提供了依据。The structure of prestressed steel structure is complex,and the safety performance involves various influencing factors,design parameters and mechanical parameters.In order to realize the intelligent evaluation of the safety state of prestressed steel structure,an intelligent evaluation method for the structural safety performance based on multi-source data was proposed.In the research process,the evaluation was divided into two aspects:analysis and prediction of structural safety performance.Firstly,the relationship between multi-source data and structural safety performance was established.D-S evidence theory and BP neural network were used to analyze and predict the structure safety.On this basis,the intelligent evaluation model of structural safety was built by integrating digital twins and the implementation process was formed.Taking the chord-supported beam roof(symmetric structure)as the test object,the change of structural safety performance under temperature was analyzed,and the corresponding safety performance of different design parameters under load was predicted.Through the safety assessment method,the key stress components of the structure and the correlation mechanism between design parameters and mechanical parameters were obtained,which provides the basis for structural safety maintenance.

关 键 词：多源数据 预应力钢结构 D-S证据理论 BP神经网络 安全性能 智能化评估 

分 类 号：TU318[建筑科学—结构工程]