基于复合多尺度等概率符号化样本熵的两相流动态特性分析

Dynamic behavior analysis of two-phase flow based on composite multiscale equiprobable symbolic sample entropy

多尺度样本熵(MSE)在两相流动态特性分析中存在两点不足:一是熵值无法单纯反映时间序列信息增长速率,在高尺度下稳定性较差;二是传统粗粒化过程中有部分数据信息丢失.针对上述问题,提出复合多尺度等概率符号化样本熵(CMESSE),并通过对几种典型非线性时间序列进行分析验证了其有效性.与MSE相比,CMESSE不仅能够有效表征不同动力系统非线性时间序列复杂性,而且在时间序列较短时稳定性更好.在此基础上分析了123组流动条件下垂直上升管内空气-水两相流压差波动时间序列.研究结果表明,泡状流、塞状流及混状流的CMESSE变化趋势能够在不同尺度下定性表征不同流型的动态特性,CMESSE复杂性指数可跨多尺度定量描述不同流型的动力学复杂性.

Multiscale sample entropy(MSE)in dynamic behavior analysis of two-phase flow suffers from two shortcomings:One is that entropy value does not purely characterize the time series information generation rate,which makes the stability poor in high scale;The other is that the traditional coarse-graining process leads to some data information loss.For the above problems,composite multiscale equiprobable symbolic sample entropy(CMESSE)is proposed.Through the analysis of several typical nonlinear time series,the effectiveness of CMESSE is verified.Comparing with MSE,CMESSE not only effectively characterizes the complexity of nonlinear time series for different dynamic systems,but also shows better stability in short time series.Based on this,differential pressure fluctuation time series of air-water two-phase flow under 123 flow conditions in vertical upward tube are analyzed.The results show that the variation trend of CMESSE of bubble flow,slug flow and churn flow can qualitatively characterize the dynamic behavior of different flow patterns at different scales.The complexity index of CMESSE can quantitatively describe the dynamical complexity of different flow patterns across multiple scales.