...
  • 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>IEEE sensors journal >Improving the Detection of Explosives in a MOX Chemical Sensors Array With LSTM Networks
【24h】

Improving the Detection of Explosives in a MOX Chemical Sensors Array With LSTM Networks

机译：用LSTM网络改善MOX化学传感器阵列中爆炸物的检测

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Entities throughout the world face the problem of detecting hidden explosives, where human and canine inspection might not be a viable solution. Therefore, it is important to develop fast, reliable, and portable integrated inspection systems by means of automated methods, such as electronic noses. The goal of the work presented here is to develop an accurate, fast and light-weight machine/deep learning classification model to be used in a MOX chemical sensors array (electronic nose), in order to identify explosive substances. For this paper, 140 samples were taken, combining TNT or gunpowder with either soap or toothpaste, or acquiring raw samples of those substances in amounts ranging from 0.1 g to 2 g. For the classification problem, among the different options in machine learning techniques, five models were evaluated. The implemented LSTM version of a LeNet-5 based network, classifies accurately the compounds in 100% of the cases when using only 30 seconds from the 360 obtained by the sensor array per each sample. The results of this work indicate that the proposed LSTM-based deep learning model could be easily implemented into an embedded system.
机译：世界各地的实体面临着检测隐藏炸药的问题，人类和犬类检查可能不是可行的解决方案。因此，重要的是通过自动化方法开发快速，可靠和便携式的集成检查系统，例如电子鼻子。这里提出的工作的目标是开发一种准确，快速的轻质机器/深度学习分类模型，可用于MOX化学传感器阵列（电子鼻子），以识别爆炸性物质。对于本文，采用140个样品，将TNT或火药与肥皂或牙膏组合，或以0.1g至2g的量获取那些物质的原料样品。对于分类问题，在机器学习技术的不同选项中，评估了五种模型。基于Lenet-5的网络的实现的LSTM版本，在每次样本的传感器阵列所获得的360中仅使用30秒的情况下，在100％的情况下准确地分类化合物。这项工作的结果表明，所提出的基于LSTM的深度学习模型可以很容易地实现为嵌入式系统。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号