| Scientists have pay attention to the measurement of water pipe leak, which has brought a series of serious aftereffects including the daily living, work and industry. The advance technology of pipe detection: SIM (structural integrity monitoring) system is studied in this paper. Aimed at the SIM system, this paper emphases on the modelling the pipe deterioration and the acoustic signal detection. Fuzzy-based and half markov model is presented in the modelling the pipe deterionation, the future information can be forecasted based on the model. In the pipe leak detection, the characterstics of the buried pipe leak signal is analysied firstly, then the denoising method based on partial translation invariance is presented, thirdly, the leak signals are recognised by ChiMerge rough set, finally, the leak point pinpoint method of combination of wavelet and EMD is presented. The main works and innovation methods are listed as follows:(1) A key question that when to detect the water pipe is presented because of the few appearance of water pipe leak. Aimed at the situation, fuzzy-based and half markov model is presented. The pipe age, situation and degree of deterioration are de-fuzzied firstly, then the buried water pipe situation transfer matrix is set up, the future situation of the buried pipe can be forecasted based on the transfer matrix. The virtue of the model is that the future information can be forecasted only based on the presently information. The optimization detection time can be determined by this information.(2) The buried pipe transfer equation, the frequency character and the attenuation equation of the leak signal are studied. Questions that the attenuation of leak signal and whether the leak signal can be detected exists based on acoustic signal detection. The conclusion that the frequency of leak signal is below 1000 Hz is achieved by discussing the transfer equation and the analysis of the leak signal, then the attenuation equation of S=l wave is found, the later conclusion proves that the leak signal can be detected based on acoustic sensors.(3) The buried pipe leak signal is denoised based on the method that the partial translation invariance is studied, and the new evaluation standard (cross-correlation) of the effect of the buried water pipe leak signal denoising is produed. De-noising based on wavelet transform may bring artifacts (Gibbs) on singularity of signal. Because of great computation work of the method-trabslation invariance that can eliminate Gibbs, the partial translation invariance is presented. We firstly knowthat Haar wavelet can eliminate Gibbs if the singularity is in the middle of signal by analysis of the Gibbs, then the method finding the location of singularity is produced, finally the partial translation invariance is descried in detail. A new evaluation standard of the effect of the buried water pipe leak signal denoising based on practical situation is also presented. Simulation and practical experiment indicates that partial translation invariance can achieve good effect while reducing translation work, and also can improve the precision of the water pipe leak location.(4) Aimed at the recognition methods that can only recognise, and cannot predigest the character of leak signal, a method of leak recognition based on acoustic sensor and ChiMerge rough set theory is produced. This method is based on the data sampled by acoustic sensors, and the decision table is establisted by combining the EMD and feature of power extraction, then the decision table is discretized by ChiMerge arithmetic, finally the buried water pipe leak is recognized by reduction of rough set theory. Experiments show that the method not only recognizes the leak signal, but also can achieve the important frequency sects by reduction table, that can offer the proof for the later leak location.(5) Aimed at the disadvantages of direct cross-correlation method to pinpoint the leak point, methods based on wavelet, EMD and the fusion of wavelet and EMD are put forward to locate the leak point. The EMD method is that the signal are decomposed into a series of IMF, each of which covers a specific frequency band, then select a few useful IMFs by feature extraction of power, and locate the water pipe leak using the method of cross-correlation. The wavelet method is that the signal is decomposed into a series of frequency band, a few useful frequency bands are selected by power feature extraction, and the leak point is determined by comparing the correlation coefficient. The fusing method is that fusing of wavelet and EMD, and achieves the best effective results.Keyword: buried pipe, half markov model, partial translation invariance, ChiMerge rough set, wavelet packet decomposion, EMD, SIM system...
|
PDF Full Text Request