Study On Signal Analyzing And Processing For Leak Detection And Location In Water Pipelines

The leakage of pipe is one of the emergency problems confronted by water supply industry. It is necessary to find and locate the leak occurrences timely, which can be beneficial to maintain the safely running of pipeline and avoid the waste of water resource. The research on the leak detection and location of pressure pipelines mainly focuses on the long-distance oil and gas pipelines. Although the techniques involved in a leakage control program of long-distance oil and gas pipelines are well developed, the techniques are almost not appropriate for using in water pipelines. The reasons are as follows: 1) Because of the size and complexity of the water distribution network, the major and minor pipes in the network form a topological structure with multi-branch and multi-node. 2) Because of the different supply arrangements between major and minor pipes, the pipe conditions in the distribution network are diverse, such as the pipe size, pipe material, the pipe wall thickness, the joints and the cover conditions. 3) Leakage occurs in different components of the distribution system: transmission pipes, distribution pipes, service connection pipes, joints, valves, and fire hydrants. Causes of leaks include corrosion, material defects, faulty installation, excessive water pressure, water hammer, ground movement due to drought or freezing, and excessive loads and vibration from road traffic. Thus, the developed methods for leak detection and location in the long-distance oil and gas pipelines are almost not suitable for application in the water pipelines.Acoustic leak detection techniques have been shown to be effective and are widely used in the nondestructive testing of water industry. The current studies mainly focus on the leak acoustic signal propagation characteristics and the optimization of leak location methods based on time delay estimation. In this thesis, the mechanism of leak acoustic signal generation, the feature extraction and identification and a new leak location method are investigated. The main work is provided as follows:1. Aiming at the mechanism of leak acoustic signal generation, we investigate the leak fluid field in pipe by computational fluid dynamics (CFD). The results show that the unsteady flow separation at the leak hole, cavitations, and fluctuating pressure in a turbulent boundary layer are the excitation sources of leak noise. The effects of leak type, pipe pressure on the excitation sources are investigated. The vibration characteristics of the pipe under excitation are analyzed based on the thin shell vibration principle. The experimental results show the validness of the mechanism analyses.2. According to the generation mechanism of the leak acoustic signal, the characteristics of the leak acoustic signal are investigated. The auto correlation technology is adopted to descript the leak signal characteristics due to the ability to analyze the coherence of time series. A new procedure to identify the leak acoustic signal from the disturbed noises is proposed based on the conjunction of correlation and approximate entropy algorithm.3. The principle and application in leak location of the correlation and the adaptive filter techniques are introduced. The problems of the traditional leak location methods are analyzed.4. In order to establish the leak detection signal model, the leak acoustic propagation characteristics are investigated experimentally. The leak acoustic channels may be supposed to be linear time-invariant FIR systems. It is noted that a leak acoustic channel system has an impulse response with quite a long sequence. Because the overlap-save technique is a block processing where the channel coefficients being identified are updated once per block of input data, it is accurate for the identification of a long impulse response. The cross-correlation fitting technique shows superior performance when the channel is close to unidentifiable. Therefore, two cost functions are constructed based on the overlap-save and the cross-correlation fitting techniques to identify the leak acoustic channels. As genetic algorithm (GA) requires no gradient calculation and is less susceptible to local optima, we adopt GA to optimize the cost functions in blind channel identification. The practical identification results show the validness of the proposed scheme.5. Since the collected signals are heavily blurred by bursting interferences and present non-stationary, we propose a method to discriminate and remove the burst-type noise. Actually the acquired signals contain the characteristics related to the acoustic propagation channels, thus the blind system identification strategy is applied to estimate the transmission performances of acoustic channels. Then the times due to the propagation of the leak source signal traveling from the leak point to sensors are determined. The leak noise propagation velocity is directly calculated based on the estimated propagation times and pipe length. Here the variations of propagation velocity with pipe conditions are investigated experimentally. Due to the fact that the propagation times can be estimated by leak acoustic channel identification, for leak location, even if the information of pipe conditions is not available, the leak point can be located. This way, for leak location, the detection distance or the propagation velocity is no longer a prerequisite.6. The intelligent leak detection and location instrument system consists of a host computer, multiple data acquisition units connected to accelerometers integrated with ICs. Based on the virtual instrumentation, the hardware combined with the software implements the functionalities of information acquisition, storage, transmission, processing and presentation.