Flow Patterns Identification With Information Entropy

Based on existing fractal and chaotic theory, nonlinear theory was used to anlysis conductance fluctuating signals of oil-water two phase flow in vertical upward flow pipes for 51% ≤Kw ≤ 91% and 10 m3. d-1 ≤Q, ≤60 m3. d-1 in this paper . The information entropy is calculated and its variation is discussed . The problem which identificate oil-water two phase flow pattern with information entropy is discussed profoundly. The information entropy distribute in the range from 0.1209 to 0.1659 for 61% ≤Kw ≤ 91% .Changes of the information entropy with variations of total flowrate and water- cut correspond to oil in water in existing flow pattern map. The information entropy distribute in the range from to 0. 1420 to 0. 1844 for Kw=51% or Kw =51.5%, and the show an irregular sudden change with variations of total Flowrate Q, , which correspings to happenings of transition flow pattern map. It is shown that the information entropy is sensitive to flow pattern variation as "indicator" . There are many difference between information entropy and fractal ,but they are same on oil-water flow pattern variation , therefore, that identificate flow pattern with information entropy put forward a new method to identificate flow patterns.In the light of the time series chosen is chaotic time series , crucial problem is how to define reasonable time delay . Solutions in detail was left out in many references. The time that mutual information first go to minimum is chosen as the time delay by compiling Visual Basic programs .Which is a new probe.Correlation dimension and information entropy for the time series is calculated at same time with a Visual Basic program which adapt to deal with chaotic problems.