摘要：Traditional sample entropy is greatly affected by the tolerance r, which is determined by the original data standard deviation. Entropy values are subject to the non-stationary mutations and the probability distributions of sequences, which cannot reflect the growth rate of new information purely. To solve problems above, we combine equiprobable symbolization with the sample entropy and detail the practical significance of equiprobable symbolization sample entropy (ESSE), mathematical model and its parameter selection. And the presented method is applied to the signals analysis of gas liquid two-phase flow. From the results, the ESSE can not only identify the different flow patterns, but also can reflect the dynamic evolution characteristics of flow patterns. Compared with the traditional sample entropy, the ESSE method can resist the interference of non-stationary mutation more effectively, and obtain dynamic characteristics of the short sequences quickly and accurately. The results show that the ESSE is of significance in flow pattern identification and evolution feature analysis of two-phase flow signals. © 2016 TCCT.

卷号：2016-August

是否译文：否