Study On Analysis Method Of Dynamic Fluctuating Signal Of Gas/Liquid Two-Phase Flow

In this paper, the conductance fluctuating signals of gas/liquid two-phase flow are disposed by using Probability Density Function, Power Spectrum Density, Short Time Fourier Transition, Wigner-Ville Distribution and Hilbert-Huang Transform. The results of the study indicate:The dynamic fluctuating signals of five typical flow patterns are disposed by Probability Density Function and Power Spectrum Density analysis methods, and the distributing characters of the results are not obviously different, so Probability Density Function and Power Spectrum Density analysis methods can to be used to distinguish the flow patterns. Short Time Fourier Transition is a time-frequency region analysis method which is simple to account. A good result is obtained when this method is used to analyze the dynamic fluctuate signal of slug flow, which has legible intermittence distributing character. But it can't distinguish other flow patterns. Wigner-Ville Distribution is one of the most popular signals analysis methods in time & frequency domain, which is a powerful implement for analyzing the non-station signals. From the analysis result, bubble flow, slug flow and churn flow are distinguished preferably. It's difficult to distinguish transition flow patterns.Hilbert-Huang Transition is based on Empirical Mode Decomposition and has the character of self-adapting for signals. In this paper, Wigner-Ville Distribution and Hilbert-Huang Transition are combined for analyzing the conductance fluctuating signals. And it is easy to distinguish five typical flow patterns: bubble flow, bubble-slug flow, slug flow, slug-churn flow and churn flow. Based on the results, the flow patterns of every signal is defined accurately, and then the physics model of the transition of gas/liquid two-phase flow is contrasted which has been published. And from the view of distinguishing the flow patterns using experiment measurement signals, it proves that the model is not conformable for this experiment adopting big diameter pipe.Moreover, the Hilbert-Huang Transition spectrum is obtained based on Hilbert-Huang Transition, and the energy & frequency characters are picked up. The main components of the original signals are found after studying the characteristic of Intrinsic Mode Functions in time & frequency domain. The result of identifying the flow patterns is satisfactory based on neural network forecast model and those energy & frequency characters.