The study of dynamic differential pressure signal of gas-liquid two-phase flow based on adaptive Chirplet transformation

The gas-liquid two-phase flow shows non-stationary characteristics which cannot be well analyzed by traditional frequency or time domain analysis.In this study,the dynamic differ-ential pressure signal of two-phase flow is sampled for flow characteristics analysis and flow pat-tern identification.A time-frequency analysis technique with adaptive Chirplet transformation is introduced,which could profile the energy intensity and density distribution of non-stationary signal.The multi-hole orifices plate flowmeter is used to generate the dynamic differential pres-sure signal in a horizontal gas-liquid two-phase pipe.The dynamic differential pressure signals sampled in the pipe are of non-stationary characteristics which are determined by the flow patterns.The adaptive Chirplet transformation is adopted to process the dynamic differential pressure signals.The stages of the method are:Firstly,the differential signals are decomposed and reconstructed by optimized parameters and the ratio of the residual energy of the signal to the total energy is kept lower than 10% to ensure the useful components of the signal are completely decomposed.Secondly,the time-frequency distribution characteristics of different flow patterns are analyzed by transforming the reconstructed signals.The time-frequency spectra of the trans-formation contain signal components’time-frequency information which reflects the flow proper-ties of different flow patterns.The time-frequency spectra show that the energy of bubbly flow is mainly concentrated in a relatively high-frequency band (15-35)Hz with a low energy intensity. The energy of slug flow is distributed in the low-frequency band (0-5)Hz and the high-frequency band (10-35)Hz,and the signal energy is increased.The energy of plug flow is of low frequency characteristic ranging (0-5)Hz,and the energy intensity is the largest among the three flow pat-terns.Thirdly,three eigenvalues from the spectra are extracted and a 3D scatter plot chart is established,aiming to connect the signal features with flow pattern characteristics.It is found that there is a connection between the flow variances and the alteration of eigenvalues,which could identify the gas-liquid two-phase flow patterns.This method,having higher time-frequency resolution,could provide good noise reduction performance.The instantaneous frequency infor-mation of the signal components could be clearly demonstrated.