西安建筑科技大学学报(自然科学版)

选取与目标设计谱匹配的地震动记录是建筑结构抗震分析与设计中的重要问题之一。鉴于卷积神经网络在图像识别与分类领域具有突出的性能,本文提出了一种基于卷积神经网络的谱匹配地震动选取方法以考虑反应谱图像的二维特征。首先阐述了通过卷积神经网络进行谱匹配地震动选取的基本原理和训练数据的生成方法,然后讨论了Keras深度学习框架中11个神经网络结构在谱匹配地震动选取中的特性,最后采用Xception和InceptionResNetV2两种性能最优的网络结构从200次地震的11462条地震动水平分量中选取与我国规范设计谱相匹配的数据。研究表明,卷积神经网络能够有效地提取反应谱图像的特征,所选地震动平均反应谱与目标设计谱之间的差异较小。本文的研究工作可为工程实践中选取与目标设计谱匹配的地震动提供方法参考和技术支撑。

Selecting the spectrum-matched ground motions is one of the critical problems in the seismic analysis and design of building structures. Because of the outstanding performance of convolutional neural networks in image recognition and classification, this paper proposes a spectrum-matched ground motionselection methodbased on convolutional neural networkto consider the two-dimensional characteristics of response spectrum images. Firstly, the basic principles of using convolutional neural networks to select spectrum-matched ground motions and the method of generating training data are described. Then, the characteristics of 11 neural network structures of the Keras deep learning framework in the spectrum-matched ground motions selection are discussed. Finally, two optimal networks of Xception and InceptionResNetV2 are used to select the records matching the design spectrum of the Chinese design code from 11462 horizontal components of 200 earthquakes.The results show that convolutional neural networks can efficiently extract the features of the response spectrum image, and the difference between the mean response spectrum of the selected motions and the target design spectrum is non-significant. The research work could provide method reference and technical support in engineering practice when the spectrum-matched ground motion selection is needed.

引言
1 基于卷积神经网络谱进行谱匹配地震选取的基本原理
2 不同卷积神经网络结构性能分析
3 基于卷积神经网络的地震动选取
4 结论

图1 基于卷积神经网络进行谱匹配地震动选取示意图<br/>Fig.1 Schematic diagram of spectrum matching ground motion selection based on convolutional neural networks

图1 基于卷积神经网络进行谱匹配地震动选取示意图
Fig.1 Schematic diagram of spectrum matching ground motion selection based on convolutional neural networks

图2 本文所选记录的震级相对于震中距的散点分布图<br/>Fig.2 Scatterplot of magnitude relative to epicentral distance for the ground motions selected in this paper

图2 本文所选记录的震级相对于震中距的散点分布图
Fig.2 Scatterplot of magnitude relative to epicentral distance for the ground motions selected in this paper

图3 本文所用卷积神经网络的训练样本示意图<br/>Fig.3 Schematic diagram of the training sample for the convolutional neural networks used in this paper

图3 本文所用卷积神经网络的训练样本示意图
Fig.3 Schematic diagram of the training sample for the convolutional neural networks used in this paper

图4 本文所选11种卷积神经网训练集准确率与损失函数值随迭代次数的变化曲线<br/>Fig.4 Curves of the accuracy and loss function valuesof the training set to the iterations for the 11 convolutional neural networks selectedin this paper

图4 本文所选11种卷积神经网训练集准确率与损失函数值随迭代次数的变化曲线
Fig.4 Curves of the accuracy and loss function valuesof the training set to the iterations for the 11 convolutional neural networks selectedin this paper

图5 本文所选11种卷积神经网测试集准确率与损失函数值随迭代次数的变化曲线<br/>Fig.5 Curves of the accuracy and loss function valuesof the testing set to the iterationsfor the 11 convolutional neural networks selectedin this paper

图5 本文所选11种卷积神经网测试集准确率与损失函数值随迭代次数的变化曲线
Fig.5 Curves of the accuracy and loss function valuesof the testing set to the iterationsfor the 11 convolutional neural networks selectedin this paper

图6 Xception和InceptionResNetV2训练集与测试集准确率和损失函数值随迭代次数的变化曲线<br/>Fig.6 Curves of the accuracy and loss function values of the training andtesting sets of Xception and InceptionResNetV2 networks to the iterations

图6 Xception和InceptionResNetV2训练集与测试集准确率和损失函数值随迭代次数的变化曲线
Fig.6 Curves of the accuracy and loss function values of the training andtesting sets of Xception and InceptionResNetV2 networks to the iterations

图7 本文所选地震动加速度反应谱与不同Tg对应的目标设计谱的对比<br/>Fig.7 Comparisonofthe response spectra of the ground motions selected in this paper withthe target design spectrum corresponding to different Tg

图7 本文所选地震动加速度反应谱与不同Tg对应的目标设计谱的对比
Fig.7 Comparisonofthe response spectra of the ground motions selected in this paper withthe target design spectrum corresponding to different Tg

[1] LIU Y, HUANG Q, GUO Z, et al. Modal-based ground motion selection procedure for nonlinear response time history analysis of dual wall-frame structures[J]. Bulletin of Earthquake Engineering, 2022, 20(2): 765-793.
[2]曲哲, 叶列平, 潘鹏. 建筑结构弹塑性时程分析中地震动记录选取方法的比较研究[J]. 土木工程学报, 2011, 44(7): 10-21.
[3]杨溥, 李英民, 赖明. 结构时程分析法输入地震波的选择控制指标[J]. 土木工程学报, 2000, 33(6): 33-37.
[4]WANG G. A ground motion selection and modification method capturing response spectrum characteristics and variability of scenario earthquakes[J]. Soil Dynamics and Earthquake Engineering, 2011, 31(4): 611-625.
[5]JAYARAM N, LIN T, BAKER J W. A computationally efficient ground-motion selection algorithm for matching a target response spectrum mean and variance[J]. Earthquake Spectra, 2011, 27(3): 797-815.
[6]BAKER J W. Conditional mean spectrum: Tool for ground-motion selection[J]. Journal of Structural Engineering, 2011, 137(3): 322-331.
[7]LIN T, HASELTON C B, BAKER J W. Conditional spectrum-based ground motion selection. Part I: hazard consistency for risk-based assessments[J]. Earthquake Engineering & Structural Dynamics, 2013, 42(12): 1847-1865.
[8]宋亚澜, 周颖. 采用结构等效周期的强地震动记录选取方法[J]. 结构工程师, 2017, 33(3): 80-87.
[9]李琳, 任叶飞, 温瑞智, 等. 用于强震动记录选取的 EP4 均值目标谱的确定[J]. 地震研究, 2019, 42(7): 584-593.
[10]侯红梅, 刘文锋. 基于反应谱的地震动备选库建构方法[J]. 振动与冲击, 2019, 38(9): 44-52.
[11]LIU Y, KUANG J S, YUEN T Y P. Modal-based ground motion selection procedure for nonlinear response time history analysis of high-rise buildings[J]. Earthquake Engineering & Structural Dynamics, 2020, 49(1): 95-110.
[12]杨真真, 匡楠, 范露, 等. 基于卷积神经网络的图像分类算法综述[J]. 信号处理, 2018, 34(12): 1474-1489.
[13]俞颂华. 卷积神经网络的发展与应用综述[J]. 信息通信, 2019, 2: 39-43.
[14]中国建筑科学研究院.建筑抗震设计规范:GB50011-2010[S]. 北京:中国建筑工业出版社,2016.
[15]MUKHERJEE S, GUPTA V K. Wavelet-based generation of spectrum-compatible time-histories[J]. Soil dynamics and Earthquake Engineering, 2002, 22(9-12): 799-804.
[16]GéRON A. Hands-on machine learning with Scikit-Learn, Keras, and TensorFlow: Concepts, tools, and techniques to build intelligent systems[M].Sebastopol, USA: O'Reilly Media, Inc., 2019.

备注

引言

1 基于卷积神经网络谱进行谱匹配地震选取的基本原理

2 不同卷积神经网络结构性能分析

3 基于卷积神经网络的地震动选取

4 结论

期刊信息