基于遥感数据的工程结构损伤动态识别研究
Dynamic Damage Identification of Engineering Structures Based on Remote Sensing Data
关宏洁 1田晶 2王群1
作者信息
- 1. 西安欧亚学院,陕西 西安 710065
- 2. 西安市建筑设计研究院有限公司,陕西 西安 710054
- 折叠
摘要
以定期监测工程结构状况为目标,提出了基于遥感数据的工程结构损伤动态识别方法,解决获取影像粗糙问题,提升微小工程结构损伤识别效果.运用二维Tsallis交叉熵计算工程结构遥感影像的二维直方图,通过在布谷鸟搜索算法(CS)中引入Logistic映射的混沌扰动算子,形成混沌CS算法,完善二维Tsallis交叉熵的阈值选择过程,获取最佳阈值后,分割工程结构遥感影像,增强处理分割后的遥感影像,并将处理后的影像作为SVM分类识别模型输入,完成损伤动态识别、归类.实验结果表明,该方法获取遥感影像精度极高,分割影像清晰且各区域边缘完整,能够识别微小的裂缝损伤,最大程度还原裂缝宽度及线性特征,识别结果影像信息含量大,且识别均方误差低、平方相关系数高.
Abstract
Aiming at regularly monitoring the condition of engineering structures,this paper studies the dynamic damage identi-fication method of engineering structures based on remote sensing data,solves the problem of obtaining rough images,and im-proves the damage identification effect of micro engineering structures.The two-dimensional histogram of remote sensing image of engineering structure is calculated by using two-dimensional Tsalliscross entropy.The chaotic CSalgorithm is formed by intro-ducing the chaotic disturbance operator of logistic map into cuckoo search algorithm(CS).The threshold selection process of two-dimensional Tsallis cross entropy is improved.After obtaining the best threshold value,the remote sensing image of engineering structure is segmented and the segmented remote sensing image is enhanced,The processed images are input as SVM classification and recognition model to complete the dynamic identification and classification of damage.The experimental results show that the remote sensing image obtained by this method has high accuracy,the segmented image is clear and the edges of each region are complete.It can identify small crack damage and restore the crack width and linear characteristics to the greatest extent.The rec-ognition result has large image information content,low recognition mean square error and high square correlation coefficient.
关键词
遥感数据/工程结构/损伤动态识别/混沌CS算法/SVM/遥感分割Key words
Remote Sensing Data/Engineering Structure/Damage Dynamic Identification/Chaos Cs Algorithm/SVM/Remote Sensing Segmentation引用本文复制引用
基金项目
陕西省新工科研究与实践项目(陕教[2020]75号)
出版年
2024
NETL
NSTL
维普
万方数据