Feature Extraction and Analysis of Earthquake Motion
Feature Extraction and Analysis of Earthquake Motion
复制标题
地震运动特征提取与分析
DOI：
10.1007/978-981-19-4070-5_15
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
N.
中科院分区：
文献类型：
--
作者：
Kuse;M. and Nojima;N.
地震运动评价是地震工程中的基本问题。深入了解观测记录中包含的振幅、周期和持续时间等地震运动特征至关重要。本章对地震运动所包含的能够解释工程特性的特征量提出了建议。首先，提出了基于Husid图的代表强运动时间特征的99维特征向量。 Husid 图定义为标准化为总功率的地震累积功率的时间历史。此外，提出了将高斯混合模型应用于99维特征向量。高斯混合模型以多个高斯分布的混合来近似原始数据。接下来，定义了地震运动的时频分析，即基于演化功率谱的周期相关特征向量，与位移响应包络成正比。周期相关特征向量是根据归一化到总功率的进化功率谱的累积功率的时间历史计算的，参考Husid图。此外，高斯混合模型应用于周期相关特征向量。
The evaluation of earthquake motion is basic problem in the earthquake engineering. Deep understanding of the earthquake motion characteristics of the amplitude, period and duration included in observation records is essential. In this chapter, the feature quantity that can explain the engineering characteristics included in the earthquake motion, makes proposal. First, the 99-dimensional feature vector representing the temporal characteristics of a strong motion on the basis of Husid plot was proposed. The Husid plot is defined as the time history of cumulative power of earthquake motion normalized to the total power. In addition, the Gaussian mixture model was applied to the 99-dimensional feature vector was proposed. The Gaussian mixture model approximates the original data in terms of mixture of multiple Gaussian distributions. Next, the time-frequency analysis of earthquake motion, the period-dependent feature vector based on the evolutionary power spectrum, which is proportional to displacement response envelope, was defined. The period-dependent feature vector was calculated from the time history of cumulative power of evolutionary power spectrum normalized to the total power with reference to Husid plot. Furthermore, the Gaussian mixture model was applied to the period-dependent feature vector.