Research On Feature Extraction And Recognition Of Gas-liquid Two-phase Flow Regime Based On Fractal Theory And Entropy Theory

There are a large number of multi-phase flow phenomena in our daily life.As a major branch of multi-phase flow,two-phase flow generally exists in industrial production processes such as petroleum,electricity,nuclear energy,etc.,and its heat and mass transfer rate and momentum Parameters such as loss,pressure gradient,and gas phase content seriously affect the efficiency and safety of industrial production.The flow pattern and dynamic characteristics of two-phase flow are closely related to its parameters.Therefore,it is of great significance to study the characterization and extraction of two-phase flow pattern through different analysis methods.This article describes the characteristics of the flow pattern from the two perspectives of fractal theory and entropy theory.The main work is as follows:(1)According to the nonlinear characteristics of the gas-liquid two-phase flow system,this paper introduces the fractal theory,which has a significant effect in nonlinear analysis,to describe the fractal characteristics of the two-phase flow pattern signal.By comparing the advantages and disadvantages of several traditional fractal dimension methods,we adopt a grid dimension estimation method with more comprehensive performance characteristics and simple calculation.This method obtains the fractal dimension of different signals by analyzing the signals of different sampling periods.We can identify different signals according to the size of the dimension.This article takes two-phase flow system as the research object.We use the fractal characteristic parameters to characterize the fractal properties of conductancefluctuation signals of different flow patterns.The results show that the theory can effectively identify different flow pattern signals.(2)Aiming at the problem of uneven coverage of signals based on box dimension fractal theory,which affects the accuracy of dimensionality,this paper proposes a fractal dimension method based on mathematical morphology to extract the fractal features of the flow pattern signal.We use flat structure elements to explore the details and characteristics of flow-type signals,which can efficiently characterize and depict the signal,and do not need to analyze the signal in the time domain and frequency domain.This method avoids the shortage of box dimensions and the result is more accurate and the calculation is simpler.The results prove that the fractal dimension estimation method based on mathematical morphology shows good performance in the extraction and recognition of flow pattern features.(3)Based on the measurement of the transfer entropy of causality between systems,this paper proposes a multi-scale symbolic transfer entropy theory to study the dynamic characteristics of different flow patterns of gas-liquid two-phase flow.First,we use the global four-symbolization method to symbolize the selected time series.Then we perform multi-scale transfer entropy analysis on the symbolized time series to obtain the entropy distribution of different flow patterns.We characterize the characteristics of different flow patterns from the internal connections between upstream and downstream of multiple scales.This method is simple and reliable,and provides a new way for the characterization of gas-liquid two-phase flow patterns.