Signal Processing For Leak Detection And Location Of Water Distribution Pipelines

Leak detection and location play an important role in the health monitoring of pipelines. To study the detection and location techniques has its theoretical and practical values.The research background is the leak detection and location for water supply pipelines. The technology of signal processing is researched in this thesis. To solve the problem of low SNR in the case of leak detection and location, which limited the accuracy of location, we focus on the aspects of sampling, noise suppression and anti-interference estimation in order to find a way to develop the ability of weak signal acquisition and processing. In the previous work, a practical smart leak detection instrument has been developed by our group. This instrument has its advantage both in low cost and high performance compared with the like products. In this thesis, the purpose is to increase the limit of minimum detectable distance and the minimum detectable leakage volume of the instrument. Follows are the main works:①Based on the previous research done by others, the leak signal of real pipe for water supply is analyzed by the method of time-frequency spectrum and spectrum analysis. The knowledge of leak signal and interference is obtained, based on which the general model of leak signal is established.②The smart leak location system for water supply pipeline designed by our group is introduced in detail including the system feature, the component of the system, and the detection principle. Starting from the leak signal general model, the leak signal model for time-delay estimate is derived. Such signal model which is used in the whole Ph. D dissertation is different from the other signal mode assumed in the method of correlation.③In the process of designing the sampling system, the method of periodic non-uniform sampling is researched including sampling realization, sampling theory, reconstruction method, and optimized method of division of sub-frequency band for reconstruction. It is proposed that in order to reduce the reconstruction error brought by non-ideal filters, the method of weighting delay should be used to reconstruct the periodic non-uniform sampled signal and to reduce the number of the non-ideal filters. It is proved that the reconstruction of periodic non-uniform sampling signal can make the sampling more effective at lower sampling rate. The division method is proposed which simplifies the reconstruction filter under the condition of lowest sampling rate. After the analysis of the optimized division of sub-frequency band, the method of sampling at Landau rate is proposed.④The merit and demerit of present method for leak detection in pipeline is detailed. A novel method of noise cancel (EWC-LP-ANC) for present leak signal in water supply pipeline is proposed. Notice that the correlation time of leak signal and noise is different, we propose using the linear prediction based on noise whitening criterion and the method of self adapting noise cancel to divide signal and noise. The ratio of power density of noise canceled signal and original signal is obtained. The ratio is used to evaluate the efficiency of noise canceling.⑤The merit and demerit of different methods of time delay estimation for pipeline leak detection are analyzed. Based on the real model of leak signal in pipeline leak detection, a new method of self adapting time delay estimation, i.e. SDF method is proposed. This method which is based on the unbiased parameter estimation can provide a better estimation than LMS method and other improved time delay estimation methods.⑥The EWC-LP-ANC method and the SDF self adapting estimation method are used in smart leak detection for water supply pipeline. By the statistic method, the detection results of each method are used to evaluate the improvement of the detection system.