Based On The Data Segment And Identify Pipeline Leak Detection And Location Technology Research

With the development of industry such as oil and natural gas, and the popularization of water supply, pipeline transport plays more and more important roles in the national economy. At the same time, with the growth of pipeline age, pipeline corrosion and some human reasons, the pipeline leak accidents happens frequently which cause resource wasting, and the leak of dangerous matter cause environment pollution and threat to people's lives and property. Therefore, the establishment of an effective pipeline leak detection system to avoid or reduce the loss as much as possible, is of great significance.With the development of varieties of disciplines, and the advances in technology, pipeline leak detection technology gradually turns from mainly hardware-based methods to software-based methods. With the continuous deepening of research, the leak detection methods based on pattern recognition have gradually developed. Based on the locating principle of negative pressure wave, the paper adopts one of the more popular machine learning methods in recent years, which is support vector machine and is used in pipeline leak detection integrating with correlation analysis, then do research in pipeline leak detection and locating.The paper firstly studies the application of wavelet transform in denoising on the pressure data, through comparing with the filter methods on frequency domain and time domain, at last chooses one-dimensional wavelet transform to do the denoising on the pressure data. Secondly, through analyzing the locating principle of negative pressure wave and the feather of wave changing, the paper proposes classify the sub wave on the wave curve by adopting SVM. The SVM can identify the class of the sub wave and filtrate the sub wave which falls abruptly containing the information of leakage. Then according the result of SVM, the paper introduces some correlation analysis methods base on different feathers, and puts use on the pressure data the pressure sensors received to do the leak detect and leak judgement . At the same time, the paper analyses the capability of the correlation analysis method under different noise. Finally, on the basis of the locating formula of negative pressure wave, the paper discusses the factors which affect the locating, and gives a locating strategy based on statistics.