Multilayer Network from Multiple Entropies for Characterizing Gas-Liquid Nonlinear Flow Behavior

Abstract

Characterizing nonlinear dynamic behaviors underlying multiphase flow has attracted considerable attention from the nonlinear research field. In this paper, the authors develop a novel multiple entropy-based multilayer network (MEMN) for exploring the complex gas-liquid two-phase flow. At first, we carry out the gas-liquid flow experiments to get the multichannel measurements. Then, MEMN is constructed based on the fusion of three nonlinear entropies, namely weighted permutation entropy (WPE), wavelet packet energy entropy (WPEE), and amplitude entropy (AE). For each derived projection network of MEMN, spectral radius and global clustering coefficient are both calculated and they allow effectively uncovering the nonlinear flow behaviors in the transition of different gas-liquid flow patterns. In addition, we perform wavelet time-frequency representation for the two typical flow patterns and the results support our findings well. All these demonstrate that our MEMN framework can effectively characterize the nonlinear evolution of gas-liquid flow from the perspective of complex network theory. And this also provides a novel idea for studying nonlinear complex systems from the observed multivariate time series.