Weak Signal Stochastic Resonance Method For Ultrasonic Guided Wave Detect Pipeline Deformation

Now,the pipeline transportation is widely used in the transportation of petroleum,natural gas and other energy,plays an increasingly significant role in the development of society and economy.However,the phenomenon of pipeline,like cracks,corrosion and depressions may cause major economic losses and casualties.So,it is necessary to check the pipeline regularly.Ultrasonic guided wave detection owns a lot of advantages,such as long detection distance,fast transmission speed and full cross-section detection.It has attracted more and more attention from researchers,and has been widely applied to pipeline inspection.But the guided wave will have the peculiarities of dispersion,multimodal and attenuation during the detection process,and are affected by the noise,weak echo signal will be submerged in the noise cannot be identified.In order to extract the weak feature signal in strong noise,this paper selects a weak feature signal detection method based on stochastic resonance.The cascaded stochastic resonance models based on bistable system and Duffing system are established respectively,then put the standard ultrasonic guided wave signals with a certain noise intensity into the two stage stochastic resonance model.In order to get the maximum output signal-to-noise ratio and the best stochastic resonance effect,the adaptive stochastic resonance with parameter adjustment is first carried out,selecting the appropriate parameters according to the signal-to-noise ratio gain index.Then,ultrasonic guided wave detection is carried out for five types of deformed pipes include perfect,single depression and double depression conditions,and the numerical simulation and experimental detection signals are obtained respectively.For numerical simulation signal,it can be processed by cascaded stochastic resonance model based on bistable system and Duffing system respectively,judging the existence and position of the depression by the spectrum and time domain signals output by the system.While the experimental signal is more complex due to various factors,it needs to be processed by the combination of stochastic resonance and wavelet transform.First,the spectrum output by stochastic resonance is used todetermine whether there is a depression,then the de-noising of the wavelet transform is used to get the deformation position in the time domain.The results show that the cascaded stochastic resonance based on bistable system and Duffing system can effectively judge whether there is a depression in pipeline by whether there is characteristic frequency in the output spectrum.If there is a peak in the spectrum diagram,it shows that there is a guided wave signal in the signal,then the pipe contains the depression,otherwise it is a perfect pipe.By combining with wavelet transform,the weak feature signal can be enhanced by noise firstly,and then the residual noise is removed,so that the location of echo signal can be clearly displayed in time domain signal,thus confirming the location of depression deformation in pipeline.