The Research On Feature Extraction For Small Leak Of Oil Pipeline

As important fuel and chemical material, oil is frequently transported by pipeline, which is the most convenient and economic way for transporting. However, because of increase of service, inevitable corrosion, artificial damage and so on, leak of oil pipeline often occurs. Leak not only leads to the waste of source, but also pollutes the enviroment, even fire and explosion, which cause huge threat to lives property and living enviroment of people. These leak accidents often derive from small leak, so small leak of oil pipeline can't be ignored. Strengthening investigation and application of oil pipeline small leak detection and location technique has important meaning for reducing economic loss and enviroment pollution. In this paper, time-frequency analysis method based on vibration signal is applied to the field of pipeline small leak detection. Firstly, small leak simulation of oil pipeline is made by Fluent and Ansys, the purpose of which is to obtain vibration data of pipeline small leak. Because traditional time-frequency analysis method can't detect smal leak of pipeline effectively, a newly frequency definition and its time-freuqency distribution are presented. This freuency is named as local freuqnecy, which makes frequency definiton with more universality. As for time-frequency distribution based local frequency, its algorithm is described in detail. Through simulation analysis of combination signal, modulated signal and particular signal, the results show that concept of local frequency is correct and novel time-frequency analysis method is effective. Then, based on local frequency and EMD, the other time-frequency analysis method is presented, named as time-frequency analysis method based on local frequency and EMD, and related algorithm is given. The time-frequency distributions of simulated signals, such as combination signals, modulated signal, signal with frequency switching and particular signal, are investigated by the new approach. Compared with the existing time-frequency distribution, the simulation result shows that time-frequency distribution based on local frequency and EMD makes the frequency components display more clearly and accurately, and the time-frequency analysis approach presented is effective. Finally, vibration data derived from pipeline small leak simulation is analyzed by time-frequency analysis method presentd. Compared with Wavelet and HHT time-freqeuncy analysis method, analysis results indicate that time-frequency analysis method presented in the paper can detect pipeline small leak effectively, and offer a new and effective approach for pipeline leak detection.