The method for analyzing the flow characteristics of gas-liquid two-phase flow in vertical pipelines based on multivariate multi-scale dispersion entropy is presented.Flow pattern information of gas-liquid two-phase flow in a vertical upward pipeline is obtained using a conductivity array sensor.The dimensionality of the collected high-dimensional time series is reduced by the principal component analysis(PCA)method.Then,multivariate multiscale dispersion entropy(mvMDE)is adopted to measure the complexity of multivariate time series with different flow patterns,and compared with multiscale dispersion entropy(MDE)used for univariate time series.Further,calculate the average value of the first 10 scales of mvMDE and the growth rate of the first 5 scales.The results show that mvMDE corresponding to the same flow pattern had greater differences,and mvMDE is more sensitive to flow pattern transition.Therefore,mvMDE can more effectively reveal the evolution process of two-phase flow from bubble to slug.From the aggregation of bubbles to the gradual collapse of air plugs,from the appearance of pseudo-periods to their decline,they can all be reflected by the changes in entropy,and the joint distribution of the average value and growth rate can effectively realize flow pattern identification.
万方数据
