Fuzzy logic system to detect pump faults from motor current spectra

Faults in pumping systems can be caused through the changes in flow regimes, such as cavitation, that lead to impeller degradation or corruption of pumped material, and ultimately to the deterioration and breakdown of pumps themselves. Results are presented of experiments using spectral analysis of motor current to diagnose and predict specific failures in pumps. Fault signatures were established by relating spectral features to individual faults, and by analysing their behaviour in the presence of faults. Faults investigated were cavitation, blockage and impeller damage. A fuzzy logic system was built as a final decision making module.