Quantitative detection on internal leakage rate of pipeline valves based on background noise reduction

The detection accuracy of internal leakage rate of valves in gas pipelines is reduced for its acoustic emission detection is carried out in the complex environment with serious noise interference. In this paper, a wavelet packet soft threshold denoising method based on background noise was developed. And then, based on the denoised acoustic emission signals, the internal leakage rate of gas pipeline valves was quantitatively predicted by means of support vector regression (SVR). It is shown that when the wavelet packet soft threshold denoising method based on background noise is adopted, the obtained internal leakage source signals have less noise and the SNR of acoustic emission signals of internal leakage after noise reduction is up to 6.11. Compared with the predicted internal leakage rate of gas pipeline valves without noise reduction, the regression prediction result which is obtained by using the characteristic parameters after wavelet packet denoising is better. The prediction result of soft threshold denoising is better than that of hard threshold denoising. The average absolute ratio error of prediction result based on soft threshold denoising is 0.164. Obviously, this method improves significantly the quantitative regression prediction accuracy of internal leakage rate of valves.