| Third-party damage (TPD) activities have increasingly influenced the oil and gas transmission security in recent years. Third-party damage creates potential hazards and increased operating costs for transmission companies. TPD includes man excavate soil near pipe, drill the pipe wall, impact to pipe and boring holes, the transient waves created by impact travel in the pipe wall and inside the pipe. At present pipe leak detecting systems noted do not effectively and reliably find TPD signals, result could be significant expenditure of resources with no commensurate safety benefit. How to design a cost-effective Pipeline TPD preventive detecting system is challenging design issue. The thesis present methods to measure Third-party damage (TPD) signal in buried pipeline in real-time based on acoustic signal.According to conditions of domestic oil field, TPD mainly happen in key section(3-5km) of pump station spacing in pipeline.Construct buried pipe or cylinder wave model and analyze fundamental modes wave propagation characteristic. The wavenumbers for pipe is identified, the energy distribution among various wave types are investigated. The interest low frequency signals of TPD signals are able to propagate long distance along pipe. Using wavelet and support vector machine to deal with TPD signals. In order to maintain the preventive and real-time of monitoring system, lifting scheme of wavelet algorithm is implemented to denoise in TMS 320VC5509 DSP chip. The main works and innovative methods are listed as follows.(1) Little work about pipe TPD signal characteristic is available in the literature.Pipe equations for n=0 axisymmetric wave motion are derived for a fluid-filled pipe, surrounded by an infinite elastic medium which can support both longitudinal and shear waves. These equations are solved for two wave types, s=1, 2, which correspond to a fluid dominated wave and an axial shell wave, and expressional for a complex wavenumber for each wave are given. So we know the TPD signal attenuation characteristics in 0~ 800Hz, the low-frequency regime which is of most interest.For cast-iron, steel pipe aim signal band in 0~200Hz, if sensor span spacing larger than 200m. If sensor span less than 150m, band is 200~800Hz.(2)A modified lifting wavelet threshold denoising algorithm is used to process TPD signals and algorithm hardware implementation is perforemed in DSP chip. And, a modified cross correlation method based on wavlet packet decomposition is showed. Signals are decomposed into a series of time domain signal, each of which covered a specific frequency band to denosie and divide fine frequency band. Extract feature band to reconstruction signal.The cross correlation coefficient increases from 0.58~0.7 to 0.82~0.90.The simulation experiment in water pipe was conducted in Zhejiang University.(3)To classfy mixed TPD sounds and overcome the small amount of training sample difficulty, support vector machine is primarily applied. Extraction signal feature, divide orignal signal into 8 sub-bands then extracted 0 ~ 320Hz approximation signal is decompose into 8 bands. Each wavlet packet cover 0~ 40Hz signals. Extract packet energy as feature to classfy four TPD singals: normal condition, drilling, hammering and excavating. Detection spacing 1400m in field test are performed in crude oil pipe in Shandong Province China. Success rate is more than 85%.(4) Pipe TPD Preventive system is based on TI TMS320VC5509DSP chip, suppressing noise and spike detection algorithm is performed in DSP chip, which is implemented in fast lifting wavelet transform. Lifting scheme decreases calculation complexity and guarantee real time operation. The whole process of developping software and hareware in DSP is discussed.Industrial experiment in buried crude oil and product oil pipe show pipe wave model is correct and signals processing method is efficient, when sensor spacing is 1400m. The preventive measuring system is able to monitor the key dangerous section of pump station spacing.
|
PDF Full Text Request