| Gas pipeline is widely used in natural gas engineering,thermal power generation,chemical raw materials,pneumatic transportation and many other industrial fields,and it has many advantages such as high transportation efficiency,low safety risk,good construction economy and so on.However,due to problems such as pipeline corrosion and wear,aging and damage,geological movement and third-party destruction,gas pipeline leakage accidents frequently occur and there are serious safety hazards.In this paper,experiments and modeling studies are carried out on gas pipeline leakage based on acoustic emission technology,the characteristics of acoustic emission signals under different leakage conditions are discussed,the noise reduction of pipeline leakage signals containing solid particles is studied,a leakage source feature recognition model based on machine learning is constructed,and finally an online leakage monitoring system is developed and applied in natural gas stations.Firstly,multiple groups of pipeline leakage experiments such as different pipeline pressure,leakage pore diameter,leakage shape and non-leakage control group were carried out,and the influence of multiple leakage source characteristics on the acoustic emission signal characteristics of gas pipelines is discussed.After the pipeline leakage occurs,the energy of the acoustic emission signal increases significantly.With the increase of pipeline pressure,the total energy of the leakage signal increases,and the proportion of high-band energy increases.With the increase of leakage aperture,the total energy of leakage signal increases,and the proportion of low-frequency band energy increases.The signal energy of round,triangular and square leakage holes is close to and greater than the rectangular signal energy,and the sharpness of the edge of the leakage hole is easy to lead to an increase in signal frequency,and the signal difference between axial and circumferential leakage holes is small.Secondly,based on the improved filtering method of variational mode decomposition coupled with Spearman correlation analysis,the noise reduction of leakage signal under solid particle erosion interference is realized.Aiming at the interference of solid particle erosion on pipeline leakage signal monitoring,the acoustic emission detection experiment of mixed quartz sand was carried out,and the influence of particle size and particle concentration on the characteristics of particle noise signal is studied.It is found that the noise energy is positively correlated with particle size and concentration,and the signal frequency tends to move from high frequency to low frequency with the increase of particle size.Then,the noise reduction of the analog measurement signal is studied by the improved variational mode decomposition coupled Spearman correlation analysis filtering method.The SNR of the signal increased by 4.55dB and the MSE decreased by 0.201mV~2 under different working conditions,which means that the method achieved good noise reduction effect.Then,a method for recognizing pipeline pressure and leakage aperture based on Gaussian mixture hidden Markov model is proposed,which realizes the recognition of leakage source characteristics under pressure fluctuation interference.The acoustic emission signal of pipeline leakage under pressure fluctuation was sampled,the wavelet packet energy spectrum of the signal was extracted by wavelet packet transformation,and then the principal component analysis of the frequency band energy was carried out,and the Gaussian mixture hidden Markov model was used to realize the classification and recognition of pipeline pressure and leakage aperture.The overall accuracy of the model reaches 95.20%,the leakage aperture recognition rate reaches 99.95%,and the significant leakage recognition accuracy reaches 100%.The leakage source feature recognition can be achieved with high precision under both small and sufficient samples.Finally,the leakage online monitoring system based on acoustic emission technology is designed and developed,which consists of field detection module,server module,user interaction module and data communication module,and has been applied in natural gas stations.By analyzing the wall noise of the station pipeline,the digital filter denoising method was studied.Through the experimental study of acoustic emission propagation characteristics in pipelines,flanges,orifice flowmeters,electric ball valves and other equipment,the attenuation characteristics of acoustic emission signals after propagation through different equipment are obtained.Finally,the acoustic emission signal processing process of the online monitoring system is designed,and the on-site online monitoring client software is developed based on Qt. |
