Analysis Methods Of Multivariate Nonlinear Characters In Two-phase Flow

Two-phase flow,which is a complex flow,widely exists in scientific researches and industrial production processes.Identification of flow regimes and understanding of inherent flow behaviors are significant to the accurate measurement of two-phase flow parameters.With high request of measurement and environmental protection to industrial production,two-phase flow parameters measurement accuracy is also needed new requirements,ideas and methods.Multivariate nonlinear time series analysis methods,which regard different sensor data as a multivariate time series to describe system.Compared with single time series,multivariate time series provides more comprehensive information regarding the system's dynamics.Two-phase flow is a highly complex nonlinear dynamic system,thus can fit in different nonlinear information processing methods for flow structure and behaviour analysis.In order to characterize and identify the flow patterns of two-phase flow,the multivariate time series is extracted from an electrical resistance tomography,finding a more accurate method to distinguish flow regimes and analyze the inherent flo w behavior efficiently.The detailed research as follow:(1)Two-phase flow pattern identification,flow mechanism research and multivariable nonlinear time series methods are summarized.Due to gas-water two-phase flow and oil-water two-phase flow have chaos characteristic,the research is proposed using maximum Lyapunov exponent and correlation dimension to identify the flow pattern and analyze flow behavior.(2)The maximum Lyapunov exponent with multivariate time series can characterize and thus identify the flow pattern changes with the superficial velocity ratio of gas-water two-phase flow regime.Compared with gas-water two-phase flow,oil-water two-phase flow has more complexity.Thus,it is needed maximum Lyapunov exponent and correlation dimension to identify of flow regimes and understand of inherent flow behaviors effectively.(3)Based on the analysis of oil-water two-phase flow characteristics by multivariate time series of maximum Lyapunov exponent and correlation dimension,it can be proved chaotic characteristics of oil-water two-phase flow.The kernel partial least squares is an excellent nonlinear regression model.Compared with partial least squares,kernel partial least squares provides high accuracy prediction for complex system.Based on a 16-electrode electrical resistance tomography sensor and the soft sensor technology,a new measurement of water holdup is proposed for oil-water two-phase flow.The soft sensor technology applied in the research is a kernel partial least squares nonlinear regression algorithm.Through choice of kernel function,the computational results of four typical flow regimes of oil-water two-phase flow show that the KPLS method can real-time measure the water holdup effectively.