Study On Analysis Method Of Flow Noise Signal In Gas-liquid Two-phase Flow Tube

The two-phase flow is widespread in nature and daily life.Especially in the industrial production process,heat transfer technology and principle and gas-liquid two-phase flow are widely used in modern industry.Because of the two phase flow in the process of flow,flow characteristics and the structure of the flow will produce different flow noise,complex under different flow pattern for the study of flow noise for two-phase flow parameter measurement,actually plays a key role in the industrial system design optimization.In this paper,acoustic emission detection technology was used to investigate the flow noise in the tube.Acoustic emission detection principle is used to collect the flow noise signals of gas-liquid two-phase flow.This method is different from other two-phase flow detection methods in that it has the advantages of non-invasive,no interference to the internal flow field,and real-time online detection.In this paper,the characteristics of fluid flow in the tube are studied by acoustic emission detection system.In the time domain,frequency domain,and WVD analysis,the frictional noise between the single-phase fluid and the pipe wall comes from the low frequency band.Compared with the single-phase flow state,the two-phase flow is found to be affected by the force of the rolling suction between the two phases.The noise energy amplitude is higher and the flow process is more complicated.By extracting the time-domain characteristic parameters and wavelet energy of the gas-liquid two-phase flow noise signal in the vertical tube,and applying the analysis method of fuzzy clustering,the flow pattern can be identified online,and the recognition rate reaches 100%.Through the wavelet and R/S analysis of the flow noise of the gas-liquid two-phase flow in the vertical tube,it is concluded that the gas-liquid two-phase flow is double fractal and has both long-term and negative correlation in time series.The detail signal corresponding to the three scales of high frequency signals d1,d2 and d3 are anti-persistent,which are caused by the irregular impact between discrete phase(gas phase)and continuous phase(liquid phase)and the entrainment force caused by bubble-driven water flow.The acoustic signal received by the acoustic emission detection system corresponds to the low-band signal corresponding to the friction signal of the pipe wall,and the signal is persistent.Wavelet analysis and R/S analysis directly characterize the flow characteristics and flow mechanism of the two-phase flow in the vertical pipe,and decouple the flow noise signals in the pipe.The vertical tube two-phase flow acoustic emission signal was decomposed by wavelet and the energy values of each scale were extracted as input.The phase-containment rate measurement model was successfully established by using support vector machine regression.In the case of vertical tube flow,the maximum relative error of liquid volume fraction does not exceed 6%,and the maximum relative error of bubbly flow phase content is within 2.5%.The combination of acoustic emission and support vector machine regression techniques was proved to be successful for the establishment of the phase holdup model.