SVM Based Identification Method for Hydrate Blockage in Natural Gas Pipelines

Natural gas has been widely used around the world as a clean and environmental friendly energy. Natural gas hydrate blockage may form under a low temperature and high pressure condition, which will affect production process and even lead to severe accidents. In this paper, an identification method for gas hydrate blockage in pipeline based on acoustic excitation is presented, which can also work for pipeline leakage. An acoustic signal is transmitted into the pipeline by loudspeaker, then the reflected signals caused by hydrate blockage or leakage are acquired by a microphone near the loudspeaker. Empirical mode decomposition (EMD) method is employed to extract the eigenvectors of reflected signals, then a support vector machine (SVM) method is used to identify the types of reflected signals. Finally it is proved in the experiments that the proposed method can effective identify the hydrate blockage or leakage, and the identification accuracy rate is 97.5%.