| As a typical nonlinear dynamical system,the two-phase flows are widely encountered in the fields of energy,chemical,petrol,et al.Understanding the two-phase flow pattern evolutional dynamics is a continuous requirement of complex system research.Since there are no precisely analytic dynamical equations to model the mechanism of the two-phase fluid,revealing the two-phase flow dynamics from the observed fluid data has been widely used.However,the two-phase flow observations are easily be contaminated by environmental noise,which may lead to a misunderstanding of the two-phase flow dynamics.Especially,when the two-phase flows are shown as stationary random processes,the fluid dynamics are usually hidden behind the noises.The symbolization methods,which have been widely used in various complex systems,have been proved to be efficient tools for system dynamics detection.Since the symbolization methods are robust to noise and fast computing,it provides a new way of analyzing the twophase flow observations.In this regard,we in this paper propose a series of symbolic methods for the two-phase flow characterization,with which we investigate the gas-liquid two-phase flow pattern evolutional dynamics,the unstable periodical oscillations,and the flow pattern spatial coupling characteristics.The main contents of this paper are as follows:First,we design a gas-liquid two-phase flow experiment and measure the conductance fluctuation signals of the gas-liquid two-phase flow system.We calculate symbolic recurrence plots from these two-phase flow fluctuations and evaluate flow pattern evolution by using symbolic recurrence qualification analysis.The mechanisms of flow pattern transition from slug flow to bubble are also clarified.Then,we propose the gas-liquid two-phase flow symbolic transition network,from which the enumerated prime cycles are used to characterize the unstable periodic orbits in the vertical gas-liquid two-phase flow system.We also investigate the flow gas-liquid twophase flow state transitional characteristics with the calculated network degree variance,average degree,and average shortest path length.We also propose the multivariate coupling symbolic network,which is constructed by the multivariate four-sector sensor conductance signals,to identify the flow pattern spatial coupling dynamics.The network topology indicates that the coupling strength is gradually reduced when the flow pattern changes from slug to bubble.We calculate the mutual information mean,local weighted clustering coefficient,clustering coefficient entropyas indexes to qualify the coupling strength of different flow patterns and investigate its evolutional characteristics. |
