机器学习在简支梁桥损伤识别中的研究
投稿时间:2020-04-28  修订日期:2020-06-09  点此下载全文
引用本文:
摘要点击次数: 125
全文下载次数: 0
作者单位邮编
石敏* 贵州大学 550025
钱松荣 贵州大学 
基金项目:贵州省科技计划项目(黔科合支撑[2019]2886)
中文摘要:为了保障结构的健康,提出更好的算法以便建立结构健康的分类模型。以简支梁桥为例,有三个振幅不同的激振源和一个可变弹簧,通过位于简支梁桥上的47个传感器获得横向加速度以及相应的结构损伤程度的数据集。先通过最大相关峭度反卷积和极值统计法预处理数据集,然后通过快速傅里叶变换后再极值统计,最后使用线性支持向量机加网格搜索法准确率达到了95%。得到了一种更适合简支梁桥的损伤预测方法。
中文关键词:结构健康监测  简支梁  机器学习  最大相关峭度反卷积;快速傅里叶变换
 
The study on the damage detection of simple beam based on machine learning
Abstract:In order to ensure the health of the structure, a better algorithm is proposed to build the classification model of the structure health. Taking the simple beam bridge as an example, there are three excitation sources with different amplitudes and a variable spring. Through 47 sensors located on the simple beam bridge, the data set of lateral acceleration and corresponding structural damage degree is obtained. Firstly, the data set is preprocessed by maximum correlation kurtosis deconvolution and extremum statistics, and then extremum statistics after fast Fourier transform. Finally, the accuracy of using linear support vector machine with grid search method is 95%. A more suitable damage prediction method is obtained.
keywords:structural health monitoring  simple beam  machine learning  Maximum correlated Kurtosis deconvolution  fast Fourier transform
查看全文   查看/发表评论   下载pdf阅读器