柔性浅埋物的声-振智能探测

Acoustic-vibration intelligent detection of flexible shallow buried objects

提出一种基于目标检测算法的柔性浅埋物的声-振智能探测方法,将声波激励、激光散斑干涉测振和目标检测算法有机结合,用以柔性浅埋物的大范围快速探测。在论述YOLO系列目标检测算法原理的基础上,选择并优化柔性浅埋物的智能探测网络模型;然后,搭建声-光融合智能探测系统,构建不同柔性浅埋物的激光散斑干涉条纹图数据集;最后,对数据集进行训练和测试,验证该算法用于干涉条纹图识别的可行性。实验结果表明:在给定实验条件下,柔性浅埋物智能探测网络模型的精确率为98.39%,召回率为84.72%,平均识别精度为99.66%。该声-振智能探测方法可以在给定实验环境下对多种柔性浅埋物的激光散斑干涉条纹图进行智能识别,适用于浅层地下柔性掩埋物的大面积快速探测。

A novel sound-vibration detection approach,leveraging a target detection algorithm,merges acoustic stimulation,laser speckle interferometry,and target detection algorithms for efficient and broadrange detection of flexible,shallowly buried objects.Initially,after discussing the YOLO series target detection algorithm principles,an optimal intelligent detection network model for these objects is chosen.Subsequently,a sound-light fusion intelligent detection system is developed,creating a dataset of laser speckle interference patterns for various flexible,shallowly buried objects.This dataset is then trained and tested to evaluate the algorithm's effectiveness in recognizing interference patterns.Experimental outcomes reveal that,under specified conditions,the model achieves a 98.39%accuracy rate,an 84.72%re‑call rate,and an average recognition accuracy of 99.66%.This sound-vibration detection method effec-tively identifies laser speckle interference patterns of numerous flexible,shallowly buried objects in the tested environment,proving its efficacy for quick,large-scale detection of such objects underground.