Leak Detection And Location Of Gas Pipeline Based On Wavelet Analysis Technology

Leaks from the pipelines can not only lead to a great deal of product losses and environment pollution, but also cause many people hurt, even to die. So, leak detection and location is necessary and important for us.This study selects the negative-pressure-wave technique which is based on wavelet analysis to detect and locate the leakage for gas pipeline and pipe network. The main jobs of the paper are as follows,Multi-scale decomposition and reconstruction of wavelet transform is utilized to filter the noise of negative-pressure-wave, which is brought by leakage. With the capacity to detect the signal singularity quickly and accurately, wavelet transform was used to catch the break point of pressure signal. Accordingly, the work improved the location accuracy. For selecting the best mother wavelet and the method of thresholding procedure, the paper raises a new mean which is based on the comparison of SNR and MSE. From the example, using mother wavelet and soft thresholding procedure, we can obtain the best de-noising effect. Then, CWT to the signal with db1 and 6 scales, we can get the singularity of signal and achieve the leak location. The example is presented to illustrate the accuracy and validity of the method.On the basis of the study to the leak detection and location for single pipeline, two methods for leak detection and location of tree-like gas pipe network based on pattern recognition technology are studied in this paper. Firstly, de-noising and singularity analysis for pressure signal which colleted from each ends using wavelet analysis. Then, leak detection and location with the methods based on minimum distance classifier and neural network. The results from two methods are accordant and presented to illustrate the accuracy and validity of the methods.Enlightened by the study of leak detection and location for tree-like gas pipe network, this paper raises a new method, which is based on wavelet analysis, pattern recognition and neural network to detect and locate the leakage for ring-like pipe network. Firstly, de-noising and singularity analysis for pressure signal which colleted from each ends using wavelet analysis. Secondly, locating the preset point on the pipe network in accordance with the thought of pattern recognition and calculating the time vectors parallelisming to each ends and the time vector which would be detect; Then, using BP NN to ascertain the leak section and using Liner NN to locate the leakage singularity. The example is presented to illustrate the accuracy and validity of the method.