Hybrid fuzzy-neural flow identification methodology

In many engineering systems non-intrusive correct identification of flow regime is of great importance to performance and safety. Undesirable flow regimes may cause valves and pumps to wear, pipe leaks, even hazardous or catastrophic incidents. A non-intrusive methodology for flow identification in a two-phase flow loop is presented. The methodology relies on a fuzzy-neural hybrid system processing and interpreting impedance based measurements. The results suggest that fuzzy-neural classifiers are appropriate tools for flow regime identification.