Page 126 - 《应用声学》2022年第6期

972 2022 年 11 月

由此可知，DRSN 在对实际采样数据识别时仍 Li Feng, Wang Wenhe, You Yun, et al. Leakage detection
具有出较高的准确率，对于含噪微泄漏信号的识别， approach of gas pipeline coupling leakage noise and pres-
sure drop[J]. Journal of Applied Acoustics, 2020, 39(3):
DRSN 均能实现较好的识别效果，因此 DRSN 可用
402–408.
于对含噪泄漏信号的识别。 [7] 李俊杰, 王宏伟, 王俊生. 飞机气体系统泄漏超声信号的处理
方法 [J]. 无损检测, 2010, 32(11): 861–864.
3 结论 Li Junjie, Wang Hongwei, Wang Junsheng. Research on
ultrasonic leakage detection of military aircraft gas sys-
本文对含噪微弱泄漏的识别方法进行了研究， tem[J]. Nondestructive Testing, 2010, 32(11): 861–864.
提出以一维时域信息作为输入样本，利用 DRSN 对 [8] Wang T, Yu P, Xiao H H, et al. Detection of small gas
leaks based on neural networks and D-S evidential theory
样本进行识别的方法，通过实验验证，所得结论
using ultrasonics[J]. Insight-Non-Destructive Testing and
如下： Condition Monitoring, 2014, 56(3): 189–194.
(1) 根据泄漏上游压力不同，建立多泄漏强度 [9] 宁方立, 韩鹏程, 段爽, 等. 基于改进 CNN 的阀门泄漏超声信
数据集，验证了 DRSN 在对不同强度泄漏信号时均 号识别方法 [J]. 北京邮电大学学报, 2020, 43(3): 38–44.
Ning Fangli, Han Pengcheng, Duan Shuang, et al. Identi-
具有较强的识别性能，在对较高杂糅混合的泄漏数
fication method of valve leakage ultrasonic signal based on
据进行识别时具有较强的区分能力。 improved CNN[J]. Journal of Beijing University of Posts
(2) 利用高斯噪声模拟环境噪声，通过改变噪 and Telecommunications, 2020, 43(3): 38–44.
声含量，验证了噪声的含量并不会 DRSN 迭代次数 [10] Ning F L, Cheng Z H, Meng D, et al. Enhanced spec-
trum convolutional neural architecture: an intelligent leak
产生较明显的影响，且在对含噪泄漏信号进行识别
detection method for gas pipeline[J]. Process Safety and
时仍能保持较强的鲁棒性。 Environmental Protection, 2021, 146: 726–735.
综上所述，利用 DRSN 对噪声环境下带压气体 [11] 孙烨辰, 李鹏, 常思婕, 等. EEMD 结合改进 PCNN 模型的
管道微泄漏进行识别是一种可行的方法，同时为尽 气体泄漏信号降噪 [J]. 计算机仿真, 2020, 37(9): 409–414,
455.
可能提高含噪微泄漏识别的准确率，仍需大量实际 Sun Yechen, Li Peng, Chang Sijie, et al. EEMD com-
环境噪声样本，因此后续工作要对不同环境噪声、不 bined with improved PCNN model for noise reduction of
同泄漏类型进行分析，探讨其对泄漏识别影响。 gas leakage signal[J]. Computer Simulation, 2020, 37(9):
409–414, 455.
[12] 卢锦玲, 郭鲁豫. 基于改进深度残差收缩网络的电力系统暂态
参 考 文 献 稳定评估 [J]. 电工技术学报, 2021, 36(11): 2233–2244.
Lu Jinling, Guo Luyu. Power system transient stabil-
ity assessment based on improved deep residual shrinkage
[1] Meribout M, Khezzar L, Azzi A, et al. Leak detection
network[J]. Transactions of China Electrotechnical Soci-
systems in oil and gas fields: present trends and fu-
ety, 2021, 36(11): 2233–2244.
ture prospects[J]. Flow Measurement and Instrumenta-
[13] 车畅畅, 王华伟, 倪晓梅, 等. 基于深度残差收缩网络的滚
tion, 2020, 75(14): 1–26.
[2] Lu H, Iseley T, Behbahani S, et al. Leakage detection 动轴承故障诊断 [J]. 北京航空航天大学学报, 2021, 47(7):
techniques for oil and gas pipelines: state-of-the-art[J]. 1399–1406.
Tunnelling and Underground Space Technology, 2020, Che Changchang, Wang Huawei, Ni Xiaomei, et al. Fault
98(2020): 1–15. diagnosis of rolling bearing based on deep residual shrink-
[3] Sheen S H, Chine H T, Raptis C A. Ultrasonic techniques age network[J]. Journal of Beijing University of Aeronau-
for detecting helium leaks[J]. Sensors and Actuators B: tics and Astronautics, 2021, 47(7): 1399–1406.
Chemical, 2000, 71(3): 197–202. [14] Zhao M, Zhong S, Fu X, et al. Deep residual shrinkage
[4] Yuan F, Zeng Y, Luo R, et al. Numerical and experimen- networks for fault diagnosis[J]. IEEE Transactions on In-
tal study on the generation and propagation of negative dustrial Informatics, 2020, 16(7): 4681–4690.
wave in high-pressure gas pipeline leakage[J]. Journal of [15] He K, Zhang X, Ren S, et al. Deep residual learn-
Loss Prevention in the Process Industries, 2020, 65(2020): ing for image recognition[C]// 2016 IEEE Conference on
1–14. Computer Vision and Pattern Recognition (CVPR), USA.
[5] Evalina N, Azis H A. Implementation and design gas leak- IEEE, 2016.
age detection system using ATMega8 microcontroller[J]. [16] 廖平平, 蔡茂林, 张利剑. 气体泄漏流量非介入式超声测量方
IOP Conference Series: Materials Science and Engineer- 法 [J]. 机床与液压, 2016, 44(7): 21–25.
ing, 2020, 821(2020): 1–6. Liao Pingping, Cai Maolin, Zhang Lijian. Invasive ultra-
[6] 李凤, 王文和, 游赟, 等. 天然气管道泄漏的声压耦合识别方 sonic method for leakage flow rate measurement[J]. Ma-
法 [J]. 应用声学, 2020, 39(3): 402–408. chine Tool & Hydraulics, 2016, 44(7): 21–25.