基于长短时记忆神经网络易损性分析的适用性研究
Study on the applicability of vulnerability analysis based on long short-term memory neural network
王睿 1杨建荣1
作者信息
- 1. 昆明理工大学建筑工程学院,云南 昆明 650500
- 折叠
摘要
桥梁的损坏或失效可能导致严重的人员伤亡和巨大的经济损失.因此,对桥梁的破坏损失和地震性能进行准确的定量评估至关重要.为了实现这一目标,通常会采用构建易损性曲线的方法.易损性曲线表征在给定地震动强度下,桥梁部件或结构达到或超过某一破坏程度的条件概率.采用桥墩位移延性比作为损伤指标,利用长短时记忆(long short-term memory,简称LSTM)神经网络成功地建立了桥梁地震易损性曲线.研究结果表明,该模型展现了高计算效率和精度,可快速而准确地预测地震作用下桥梁结构构件的损伤指标.
Abstract
The damage or failure of bridges can lead to severe casualties and significant economic losses.Therefore,accurate quantitative evaluation of the damage and seismic performance of bridges is of paramount importance.To achieve this goal,the commonly adopted approach is to construct vulnerability curves.Vulnerability curves represent the conditional probability that bridge components or structures reach or exceed a certain level of damage under given seismic ground motion intensities.The displacement ductility ratio of bridge piers was used as a damage indicator,and a long short-term memory(LSTM)neural network was successfully employed to establish seismic vulnerability curves for bridges.The research findings demonstrate that the model exhibits high computational efficiency and precise accuracy,enabling fast and accurate prediction of the damage indicators of bridge components under seismic actions.
关键词
桥梁抗震/地震易损性/长短时记忆神经网络/有限元分析Key words
bridge anti-seismic/seismic vulnerability/long short-term memory neural network/finite element analysis引用本文复制引用
出版年
2024
NETL
NSTL
维普
万方数据