A CNN Model for Gas Pipeline Leakage Detection Based on MFCC Feature Extraction

被引：0

作者：

Sun, Chen ^{[1
]}

Wan, Yujie ^{[1
]}

Zhu, Peizhi ^{[1
]}

Lin, Fanqiang ^{[1
]}

机构：

[1] Chengdu Univ Technol, Chengdu, Peoples R China

来源：

PROCEEDINGS OF 2023 THE 12TH INTERNATIONAL CONFERENCE ON NETWORKS, COMMUNICATION AND COMPUTING, ICNCC 2023 | 2023年

关键词：

Convolutional neural network; MFCC; Gas pipeline; Gas leak detection;

D O I：

10.1145/3638837.3638881

中图分类号：

TP18 [人工智能理论];

学科分类号：

081104 ; 0812 ; 0835 ; 1405 ;

摘要：

The safe operation of gas pipelines in urban, industrial, agricultural, and other areas still faces risks and challenges. The pipeline leak detection method is aimed at improving the safety of gas pipeline operations in various fields and reducing losses caused by gas leaks. This method utilizes highly sensitive microphones to collect real-time sound signals in pipeline environments and uses the Mel-scale Frequency Cepstral Coefficients (MFCC) algorithm to extract features from the collected sound signals. Then, a convolutional neural network algorithm is used to identify whether there is a gas leak. This method can accurately identify pipeline gas leaks with up to 98.958% accuracy.

引用

页码：288 / 293

页数：6

共 10 条

[1]

Boski M, 2017, 2017 10TH INTERNATIONAL WORKSHOP ON MULTIDIMENSIONAL (ND) SYSTEMS (NDS)

[2] A novel pavement transverse cracks detection model using WT-CNN and STFT-CNN for smartphone data analysis [J].