Leak Detection For Based On Acoustic Emission Techniques

With the fast development of the transportation industry in the modern society, the features of high efficiency, low cost and stable supply of pipeline transportation has become one of the five basic forms of transportation nowadays. Also it has become an important part of our society. However, the pipeline leakage happens time to time because of aging, medium corrosion, abrasion, environmental erosion and being damaged by other external forces. The complex pipe laying and the limited ground conditions will make pipeline leakage hardly to be detected in time which interferes the proper functioning of the pipeline transportation seriously and it will cause great economic losses and bad social and environmental effects. Therefore, a right and timely judgment to the situation of the pipeline leakage and locating it in first time and making sure of its proper functioning has become a bran-new and great challenge to the detecting research of pipeline leakage.In this paper, it’s constructing a CO₂ pipeline leakage signal detecting platform based on acoustic emission technology. Here we are analyzing a real captured signal characteristic of pipeline leakage through theoretical analysis, data simulation and experimental verification. It presents a positioning method to pipeline leakage signal based on the combination of wavelet transformation and RBF neural network. It’s showing a research on the characteristics of burst-mode signal and continuous signal and the corresponding treating methods according to the characteristics of time domain and frequency domain of acoustic emission signal. It’s showing a research on the multi-scale decomposition algorithms and de-noise methods to wavelet transformation by aiming at the characteristics of acoustic signal under different pressures. The emphasis is on researching on the methods of extracting the vector of pipeline leakage signal characteristic based on wavelet packages. The characteristic signal of the leakage extracted by the wavelet packages is taken as the input of RBF neural network to construct the pipeline leakage positioning model based on the combination of wavelet transformation and RBF neural network to verify the validity of the models under the different pressures of leakage through the experimental data collected by the CO2 pipeline leakage detecting platform, and offering the theoretical basis for probing pipeline leakage problems and realizing fast and accurate positioning continually.