| Due to the drastic advancement of the instrumentation technology, modern signal processing and computer technology in recent years, applications of signal processing in pipeline leak detection system is becoming extremely popular. In particular, the pipeline leak detection methods based on infrasonic(low frequency) wave signals show the most accurate performance because of its slow decaying characteristic in both gas and liquid transmission.Through our continuous studies to the signal propagation behaviors in pipeline transmission, we have established a pipeline leak detection model including both alarm and location. We have first collected acoustic signals under both normal and leak conditions, and then applied various signal analysis to these two conditions in time, frequency and time-frequency domains. Our analysis and experiments show that the cross-correlation method outperforms the LMS method in time delay estimation(TDE), especially under noisy environment. Furthermore, we have also found that wavelet packet analysis is able to extract more accurate information than both wavelet analysis and the wavelet de-noise technique. Our results are also supported by onsite data collected from realistic operation environment.We have also conducted an onsite field experiment on a domestic long distance transmission pipeline to compare two TDE methods based on wavelet transform. The results once again confirm that the cross-correlation in combination with wavelet packet analysis method has the best leak location accuracy. The error in fluid velocity estimation is found to cause the most severe impact to our current method.Finally, we have developed a “pipeline leak detection system” based on the Lab VIEW program from National Instruments. Our system has extensive functionalities, high location accuracy and is convenient to operate. Our system can be also easily adapted to any oil and gas pipeline for leak detection and location. |
PDF Full Text Request