The Fundamental Research On Leak Detection For Water Pipeline System

The work of controlling leak water of pipeline system is being paid more attention to by countries all over the world. The amount of controlling leak water is an important criterion of water pipeline technology and running state. All the water-supply enterprises are faced with the same topic of controlling of leak water. At the same time, the equipment of water-supply enterprises in our country is out of date. Despite of some achievements in searching leak technology in more than ten years, they are seldom used in practical work because of dissatisfied effect or very large investment. It is an urgent task to search an economic and effective method for all these supply water enterprises. Based on the methods being used and some achievements of new technologies issued recently, some problems about the technology of leak detection are discussed in the thesis, and a new method, which is used only one sensor and based on neutral network, is put forward. It is some new attempt to use LabVIEW and MATLAB to compose programs on signal processing and pattern recognizing. The mechanism of the stress wave caused by water pipelines leak turbulence injection and its propagation in the pipeline shell are researched in the thesis, the longitudinal wave, transverse wave and circle wave are analyzed. The simulated experiment is designed. The signals from the leak of different shapes and different distances on the same pipeline are token down and processed. The analysis of the coherence between two signals in. time domain is done to examine the leak location. Simultaneously, it proves the imitated experiment is workable. Processing the signal of one sensor collecting on the different positions of water pipeline leakage using LabVIEW is researched in the thesis. A series of characteristic index are described to extract the information of stress wave in time domain and frequency domain. A back propagation neutral network (BPNN) with the input matrix composed of the characters of the stress is proposed to detect the leaks in the fluid pipelines and localize the leak. The back propagation neutral network, which classifies the state of leak detection, is selected for the model of neutral network. Training and checking are finished through MATLAB in computer. the method is verified valid by the practice. In the portion of collecting signal, because of the expensive price of AID cards in NI, a program to apply the AID cards in our country is also supplied in the thesis. Of course, it will cut down the cost greatly.