De-noising of acoustic signal by natural gas pipeline leakage based on DTCWT-SVD

摘要

The acoustic signal is generated by high-speed jet flow from inter pipelines and is used to detect leakage. The acoustic signal has broadband and chaotic characteristic, its energy concentrate on low frequency region (<;100Hz) and mixes with complex noise. Traditional de-noise methods are not suitable for acoustic signals. A novel de-noise algorithm based on dual-tree complex wavelet transform and singular value decomposition (DTCWT-SVD) is applied for leakage acoustic signal in this paper. The processing flows of de-noising is presented and optimized, which applies to remove noise from acoustic signal. Based on partial decomposition in frequency domain and local properties of leakage acoustic, the optimal decomposition level can be obtained. Phase-space reconstruction matrix is built by the time domain join method, which can decrease calculation quantity and complexity. With PCA algorithm determines the threshold of contribution rate for diagonal matrix. Using experiment data verify the effect of serval de-noise methods which include DWT, SVD and DCTWT-SVD. The test result indicates that DCTWT-SVD has a better de-noise effect.