Pressure vessel leakage detection method based on online acoustic emission signals

Pressure vessel leakages cannot initially be visited directly and will gradually cause deterioration, which can result in catastrophic damage. Acoustic emission (AE) signals generated by leakage have the potential of being used for online monitoring. Unfortunately, AE signals have the characteristics of being non-stationary, wide-band and with strong noise interference, which causes the monitoring results to have low reliability. To address the poor robustness of traditional time-domain and time-frequency domain-based monitoring methods, an online monitoring method based on adaptive singular value decomposition (ASVD) is proposed in this paper. Firstly, singular value decomposition (SVD) is used to divide the signal space into a signal subspace and a noise subspace. Experiments indicate that SVD can distinguish leakages under conditions of different pressures and variable temperature, which means that SVD is sensitive to changes in signal. Subsequently, update iteration-based ASVD algorithms are proposed for long-term online health monitoring and ASVD is shown to be successful in distinguishing the different statuses of intact, leakage and repaired. To improve the robustness of ASVD, a novel energy indicator is proposed, which can identify the status change more effectively. With the proposed methodology, an online monitoring application for pressure vessel leakage detection is expected to be achievable.

Automated summary

This paper proposes an online monitoring method based on adaptive singular value decomposition (ASVD) for detecting pressure vessel leakages. By using singular value decomposition (SVD) to divide the signal space into a signal subspace and a noise subspace, the different statuses of intact, leakage, and repaired can be successfully distinguished.