Fuzzy inference systems for efficient non-invasive on-line two-phase flow regime identification

Paper in proceeding, 2009

The identification of two-phase flow regimes that occur in heated pipes is of paramount importance for monitoring nuclear installations such as boiling water reactors. A Sugeno-type fuzzy inference system is put forward for non-invasive, on-line flow regime identification. The proposed system is particularly efficient in that it employs a single directly computable input, four outputs calculated via subtractive clustering - each corresponding to one flow regime -, and four fuzzy inference rules. Despite its simplicity, the system accomplishes accurate identification of the flow regime of sequences of images from neutron radiography videos.