102
B. P. Duong et al.
Fig. 9.5 Wavelet visualization analysis of normal signal and leakage signal of 2 mm hole and 1 mm
hole
9.5 Conclusion
Owing to the fact that burst waveform in an AE signal contains vital characteristics regarding leakage phenomenon in a water-filled pipeline system, this research
proposed a reliable methodology for detecting and isolating the transient salient
burst signal distinct to the background signal with the ECFAR algorithm. This algorithm can effectively recognize and the burst waveform and robust to the variable
length of burst waveform. The effectiveness of the ECFAR detector algorithm is
indirectly assessed through classification accuracy of multiple sizes of the leak using
the obtained characteristic features from the isolated burst signal. The experimental
results illustrate that the ECFAR detector yields high-quality results in comparison to
the wavelet-based threshold burst detection method, with classification obtain to 93%
average accuracy. Unfortunately, the authors have not employed the methodology to
locate the leak point, which will be performed in future research.
B. P. Duong et al.
Fig. 9.5 Wavelet visualization analysis of normal signal and leakage signal of 2 mm hole and 1 mm
hole
9.5 Conclusion
Owing to the fact that burst waveform in an AE signal contains vital characteristics regarding leakage phenomenon in a water-filled pipeline system, this research
proposed a reliable methodology for detecting and isolating the transient salient
burst signal distinct to the background signal with the ECFAR algorithm. This algorithm can effectively recognize and the burst waveform and robust to the variable
length of burst waveform. The effectiveness of the ECFAR detector algorithm is
indirectly assessed through classification accuracy of multiple sizes of the leak using
the obtained characteristic features from the isolated burst signal. The experimental
results illustrate that the ECFAR detector yields high-quality results in comparison to
the wavelet-based threshold burst detection method, with classification obtain to 93%
average accuracy. Unfortunately, the authors have not employed the methodology to
locate the leak point, which will be performed in future research.