摘要
在结构地震易损性研究中,传统选波方法存在结果离散性大、计算量大等问题。拉丁超立方抽样方法可以有效减少抽样数,提高抽样代表性,但将其运用于离散样本抽样时,容易出现抽样失败。为此,该文对传统的拉丁超立方抽样方法加以改进,在剔除空区格后进行排序随机选点,以替代传统的坐标分量组合选点,并将其应用于基于离散地震动强度指标的地震波抽样中,以提高易损性分析效率和选取地震波的代表性。为探究该方法对结构易损性分析的影响,对改进拉丁超立方抽样和随机抽样两种选波方法进行对比,以3个不同层数的典型RC框架结构为算例,首先选取500条地震波作为输入得到基准结构易损性曲线,随后以不同维度的改进拉丁超立方抽样方法抽取5、10、15和20条地震波,以随机抽样选波方法抽取5、10、15、20、40、60、80和100条地震波,分别得到对应的易损性曲线,并比较其与基准易损性曲线的差异性和离散性。结果表明:在同等抽样精度下,改进拉丁超立方选波相对随机选波可以减少67%~80%的计算量。
In the study of structural seismic fragility, the commonly used ground motion selection methods are characterized by large discreteness of results and a large amount of calculation. Latin hypercube sampling method can effectively reduce the number of samples and improve their representativeness, however, when it is applied to discrete samples, sampling failure may occur easily. In this work, an improved Latin hypercube sampling method was proposed to select ground motions based on discrete intensity measures to improve the efficiency of fragility analysis as well as the representativeness of selected waves. It sorted the intervals after eliminating empty intervals and randomly selected points from them to replace the traditional coordinate component combination method. To investigate its effect on the structural fragility, the improved Latin hypercube sampling method was compared with the random sampling method. Taking 3 RC frame structures with different stories as examples, 500 ground motions were first selected as the input to obtain the datum structural fragility curves. Subsequently, 5,10,15, and 20 ground motions were sampled by the improved Latin hypercube sampling method of different dimensions, while 5,10,15,20,40,60,80, and 100 ground motions were sampled by the random sampling method, and the corresponding fragility curves were obtained. The difference and discreteness between the datum fragility curves and the corresponding fragility curves were studied and compared. It is found that under the same sampling accuracy, the improved Latin hypercube sampling method can reduce 67%~80% calculation workload compared with the random sampling method.
作者
季静
陈珊珊
王雨
韩小雷
Ji Jing;Chen Shanshan;Wang Yu;Han Xiaolei(State Key Laboratory of Subtropical Building Science,South China University of Technology,Guangzhou 510640,China;School of Civil Engineering and Transportation,South China University of Technology,Guangzhou 510641,China)
出处
《土木工程学报》
EI
CSCD
北大核心
2022年第3期18-26,共9页
China Civil Engineering Journal
基金
广东省自然科学基金(2020A1515010739)。
关键词
拉丁超立方抽样
框架结构
地震易损性
选波方法
Latin hypercube sampling
frame structure
seismic fragility
ground motion selection method
