文章摘要

文章信息/Info

Title:: Research on Frequency Time Series Model Based on Engineering Application

作者:: 董娣¹; 纪金豹²; 袁美巧³; 王赞军¹; 于海阔⁴; (1.重庆市地震局,重庆 401147; 2.北京工业大学工程抗震与结构诊治北京市重点实验室,北京 100124; 3.江苏省射阳县高级中学,江苏盐城 224300; 4.北京工业大学,北京 100124)

Author(s):: DONG Di¹; JI Jinbao²; YUAN Meiqiao³; WANG Zanjun¹; YU Haikuo⁴; (1.Chongqing Earthquake Agency,Chongqing 401147,China)(2.Beijing Key Lab of Earthquake Engineering and Structural Retrofit,Beijing University of Technology,Beijing 100124,China)(3.Jiangsu Sheyang High School,Yancheng 224300,Jiangsu,China)(4.Beijing University of Technology,Beijing 100124,China)

Keywords:: Ground motion; Time-frequency characteristics; Wavelet transform; Frequency non-stationary

摘要:: 对美国NGA数据库中的3 551组地震记录按场地条件、震级、震中距进行了分组,采用一维连续小波变换得到每条记录的小波功率谱。研究任意时间处小波功率谱最大值所对应的主频率,并对每个地震动分组内的主频率值做一定时间窗内的均方根处理,分别采用线性函数、指数函数和指数三角函数模型来拟合得到的主频率随时间变化的曲线,最后分析了场地条件、震级和震中距对频率时变曲线的影响,并给出了每个分组内水平向和竖直向地震动主频率随时间变化的模型参数。结果表明:地震动主频率随时间的增大逐渐减小,但是竖直向要比水平向衰减得快一些。

Abstract:: In this paper,3 551 sets of earthquake records from the NGA database were grouped according to the site condition,the moment magnitude and the epicentral distance. And then the wavelet spectrum of each earthquake record was calculated by using one-dimension continuous wavelet transform. In order to obtain time-frequency statistical characteristics of ground motion for engineering applications,the time dependent predominant frequencies corresponding to the maximum values of wavelet spectrum at every time were considered. And the root mean square of the predominant frequencies within a short time window in every group of ground motion was computed to obtain time series curves of predominant frequency. The results show that the predominant frequency of ground motion gradually decreases with the increasing of time,and the time series of predominant frequency in vertical component seems more remarkable than that in the horizontal component. Furthermore,a linear model,together with an exponential model and an exponential trigonometric model were used to fit the time series curves of the predominant frequency. Meanwhile,the effects of site condition,moment magnitude and epicentral distance on the time-varying curves of the predominant frequency were analyzed. Finally,the fitting parameters of the time series curves of predominant frequency in horizontal and vertical directions in each group were given.