Localization of underground pipelines leaks based on a comparison of acoustic portraits

Abstract

The method of leaks localization from underground water supply system on the basis of acoustic portraits pipelines comparison was proposed in the paper. The parameters of wavelet neuron network was suggested to use as the acoustic portraits, which was taught to approximate the acoustic signal, received on the soil surface above the place pipeline stowage. Forehanded search and elimination of underground water supply leakages is an urgent task, which shows a considerable practical interest. The most promising approach to the search leaks was the passive location method, which is based on the analysis of pipelines acoustic signals. The information in this acoustic signal is sufficient to make a conclusion on the present or absence of leaks. The author of previous research described the other way to search leaks, which is based on the signals’ classification with the help of artificial neuron network. The significant drawback of this approach is the necessity to study ANN (artificial neuron network), and it demands a great scope of training data and time consuming. There is an alternative way in this research, namely the leakage search method on the basis of acoustic portrait comparison. The acoustic portrait is the set of parameters, characterized the state of pipeline. The research results have shown that wavelet neuron network can be used as the acoustic portrait. The simplest criteria, the distance between parameter vectors, was used during the research to compare the acoustic portraits. The leak search method described in this paper is characterized by low computational complexity and requires a small amount of training data and that is the main advantage. The subject of future researches should be the development of more effective acoustic portrait matching criteria.