| Gas-liquid two-phase flow is a typical energy flow phenomenon in industrial production,which widely exists in fields such as natural gas pipeline transportation,nuclear reactor cooling,evaporation and condensation.Accurate measurement of flow parameters is an important basis for reliable measurement of energy,production process monitoring and process equipment design,and it is of great scientific value for further revealing the mechanism of two-phase flow.The gas-liquid flow is characterized by nonlinear and strong coupling,which brings great challenges to the accurate measurement.Based on the analysis of the interaction between light quantum(photon)and gas and liquid medium,this paper uses(Near-infrared)NIR,very high frequency electromagnetic wave and multi-sensor fusion technologies to carry out gas-liquid two-phase flow measurement experiments,the related prediction correlations are analysed and proposed.The specific contents and innovative achievements are summarized as follows:1.Based on the analysis of the interaction mechanism between molecules and electromagnetic waves of different wavelengths.The Near-Infrared(NIR)absorption sensor and the coaxial line phase sensor are designed,which are applied to the measurement of liquid film thickness and section void fraction.The multi-source sensor system was designed by integrating Acoustic emission(AE),NIR technology and Differential pressure signal,and the measurement of gas-liquid two-phase flow parameters was carried out.2.Aiming at the loss of NIR light energy caused by scattering,reflection and refraction in the process of NIR measurement,the time-frequency characteristics of signal were analyzed.Based on Variational Mode Decomposition(VMD),Sparrow Search Algorithm(SSA)and Pearson correlation coefficient(PCC),an intelligent information processing method was developed.The algorithm takes the envelope spectrum entropy as the fitness function,removes the optical interference.The details of the liquid film change are obtained,and the effective information of liquid film thickness is extractied adaptively.3.The entrance parameters affecting on liquid film thickness are analyzed,and the effects of system pressure and droplet entrainment are introduced to obtain the optimized LocktMartinelli(L-M)parameter Xmod.The liquid film thickness correlations were established by Xmod and Weber numbers(W"eg、Wel),and the mean absolute percentage error(MAPE)of the optimized correlation was 4.35%(experimental data)and 13.02%(literature data),respectively.A parameter identification method based on Particle Swarm Optimization(PSO)is proposed,which can correlate the base film thickness and wave height accurately.The MAPE of the the present model are 7.36%and 5.25%for experimnetal dada,respectively.4.After removing the baseline drift and DC component of the liquid film signal,Power Spectral density(PSD)was used to calculate the dominant frequency of the interface wave.Seven characteristic variables affecting frequency variation were extracted,the function expression of frequency prediction and Support vector regression(SVR)model of frequency prediction were established respectively.The MAPE of SVR model that trained on the data of low pressure,gas and liquid velocity is 8.18%for the high one.The MAPE of the present model are 39.93%,36.97%,29.99%,8.46%,6.07%and 7.71%for literature database,respectively.5.The void fraction was measured by coaxial line phase sensor in horizontal slug flow experiment,the literature correlations were evaluated comprehensively by experimental data and literature data.Base on constitutive equations of the one-dimensional drift flux model the new correlations of distribution parameter c0 and the drift velocity ugm are proposed respectively with the experimental data.The correlations can reflect the dynamic changes of gas and liquid velocity,density parameters and so on,and the comprehensive evaluation indexes of c0 and ugm prediction are higher than those of literature.The MAPE of new void fraction correlation are 6.98%(current data)and 11.07%(literature data),respectively,which can adapt well to the system pressure(P=0.1-0.6MPa)and pipe diameter(ID=0.019-0.101.6m).6.The horizontal slug flow parameters were collected by NIR,AE and Differential pressure multisensor system.The waveform factor and margin factor which reflect the intermittent characteristics of slug flow are extracted as the feature,the information fusion scheme is designed and the prediction effect of empirical model is evaluated.Based on Collins correlation,a model for volume void fraction and flow were established by Least absolute shrinkage and selection operator(LASSO).The MAPE of mass flow measurement model are 7.14%(mixed mass flow)and 8.75%(liquid mass flow),which realizes the measurement of gas-liquid two-phase flow non-separation. |
PDF Full Text Request