全卷积神经网络在垃圾土勘察中的应用  

Application of Full Convolution Neural Network in Garbage Soil Investigation

作　　者：徐四一 张旭 Xu Siyi;Zhang Xu(Shanghai Ease Asia Geophysical Prospecting Co.,Ltd.,Shanghai 201206,China)

机构地区：[1]上海山南勘测设计有限公司,上海201206

出　　处：《岩土工程技术》2024年第1期75-77,共3页Geotechnical Engineering Technique

摘　　要：垃圾土与原土壤往往存在电阻率差异,常用垃圾土探测方法是高密度电阻率法和时域电磁法,而对反演结果的人工解译效率低,且准确性难以保证。通过全卷积神经网络在垃圾土勘察中的应用,识别某拆后绿地改造工程地下建构筑物垃圾土探测数据,确定垃圾土范围,表明了本方法的有效性、实用性和可靠性,为垃圾土勘察、土方量计算和改善土地性状等提供参考。Due to the difference in resistivity between the garbage and the original soil,the most commonly used garbage soil detection methods are high-density resistivity method and time-domain electromagnetic method.However,the low efficiency of manual interpretation of inversion results and the difficulty in ensuring accuracy still need further study.This research introduces the application of the full convolution neural network in the garbage soil investigation.Through the identification of the garbage soil detection data of the underground buildings of a demolished green space reconstruction project,the garbage soil range was determined,which shows the effectiveness,practicability and reliability of this method.It is the reference basis for waste soil investigation,earthwork calculation and improvement of land properties.

关 键 词：全卷积神经网络 垃圾土 高密度电阻率法 异常识别 

分 类 号：P631.3[天文地球—地质矿产勘探]