Gao Yangde, Piltan Farzin, Kim Jong-Myon
Department of Electrical, Electronics and Computer Engineering, University of Ulsan, Ulsan 44610, Korea.
Sensors (Basel). 2022 May 23;22(10):3963. doi: 10.3390/s22103963.
Acoustic emission techniques are widely used to monitor industrial pipelines. Intelligent methods using acoustic emission signals can analyze acoustic waves and provide important information for leak detection and localization. To address safety and protect the operation of industrial pipelines, a novel hybrid approach based on acoustic emission signals is proposed to achieve reliable leak localization. The proposed method employs minimum entropy deconvolution using the maximization kurtosis norm of acoustic emission signals to remove noise and identify important feature signals. In addition, the damping frequency energy based on the dynamic differential equation with damping term is designed to extract important energy information, and a smooth envelope for the feature signals over time is generated. The zero crossing tracks the arrival time via the envelope changes and identifies the time difference of the acoustic waves from the two channels, each of which is installed at the end of a pipeline. Finally, the time data are combined with the velocity data to localize the leak. The proposed approach has better performance than the existing generalized cross-correlation and empirical mode decomposition combined with the generalized cross-correlation methods, providing proper leak localization in the industrial pipeline.
声发射技术广泛应用于工业管道监测。利用声发射信号的智能方法可以分析声波，并为泄漏检测和定位提供重要信息。为了保障工业管道的安全并保护其运行，提出了一种基于声发射信号的新型混合方法，以实现可靠的泄漏定位。该方法采用基于声发射信号峰度范数最大化的最小熵反卷积来去除噪声并识别重要特征信号。此外，基于带有阻尼项的动态微分方程设计了阻尼频率能量，以提取重要的能量信息，并生成特征信号随时间变化的平滑包络。过零通过包络变化跟踪到达时间，并识别来自安装在管道两端的两个通道的声波的时间差。最后，将时间数据与速度数据相结合来定位泄漏点。所提出的方法比现有的广义互相关以及经验模态分解与广义互相关相结合的方法具有更好的性能，能够在工业管道中实现准确的泄漏定位。
