期刊
IEEE TRANSACTIONS ON INDUSTRIAL INFORMATICS
卷 15, 期 12, 页码 6240-6251出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/TII.2019.2891521
关键词
Wireless sensor networks; Leak detection; Monitoring; Feature extraction; ZigBee; Informatics; Auto-correlation function (ACF); correlation coefficient; gas leak detection; weighted fusion; wireless sensor network (WSN)
类别
资金
- Key Research and Development Plan Project of Zhejiang province [2018C01036]
- National Natural Science Foundation of China [61871167, 61376117]
Industrial gas leaks cause accidents and pose threats to the environment and human life. Thus, it is essential to detect gas leaks in time. Usually, the abnormal concentration signals are defined by a fixed concentration value, such as 25 of the lower explosive limit. However, it is difficult to accumulate to the fixed point quickly when the leak is small. In addition, the actual leak signals are seldom available, making many data classifications inoperable. To solve these problems, this paper proposes a detection approach using the auto-correlation function (ACF) of the normal concentration segment. The feature of each normal segment is obtained by calculating the correlation coefficients between ACFs. According to the features of statistical analysis, a nonconcentration threshold is determined to detect the real-time signals. In addition, the weighted fusion algorithm based on the distance between the sensors and virtual leak source is used to fuse multisensory data. The proposed method has been implemented in a field by building a wireless sensor network. It is confirmed that the system detection rate reaches as high as 96.7 and the average detection time delay is less than 30s on the premise of low false alarm rate.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据