Analysis and identification of gas-liquid two-phase flow pattern based on multivariate multi-scale dispersion entropy and interconnected dispersion pattern complex network

Gas-liquid two-phase flow is polymorphic and unstable. The investigation of the flow behavior of two-phase flow is vital for the application of gas-liquid transport in ocean engineering areas. We propose an analytical framework that combines multivariate multi-scale dispersion entropy (mvMDE) and interconnected dispersion pattern complex networks (IDPCN) for flow patterns. The resistance sensor array is used to capture the two-phase flow fluctuation signals in the riser, and the dimensionally reduced time series are nonlinearly mapped into a multivariate symbolic sequence. The mvMDE of the multivariate symbolic sequence is computed to measure the complexity of the flow pattern signals at the level of amplitude variation. In addition, a new multis-cale complex network based on the symbolic sequence is proposed to study the pseudo spatial coupling behavior of flow patterns. The evolution of the flow pattern from bubble to churn flow is quantitatively characterized using the mean value of the complex network indices in the optimal scale range. Finally, we construct the joint plane of mvMDE and two network indices to complete flow pattern identification. The research results show that the framework can effectively fuse multi-channel flow pattern information at different time scales to reveal the evolution mechanism of gas-liquid two-phase flow. •Gas/liquid two-phase flow information is collected using a resistance sensor array.•The flow pattern transition mechanism is analyzed by multivariate multi-scale dispersion entropy.•The spatial coupling behavior of flow patterns is analyzed by coupling indices of interconnected dispersion pattern complex networks.•Flow pattern identification is accomplished using the joint distribution of the coupling and complexity indices and combined with the KNN algorithm.