Characterization of cavitation evolution in centrifugal pumps through motor current signal analysis

Abstract Cavitation detection is quite significant for safety and stability of centrifugal pump operation. In this research work, motor current signal analysis (MCSA) technique is improved to extract indicators for quantitative characterization of cavitation status. The method of Complete Ensemble Empirical Mode Decomposition with Adaptive Noise (CEEMDAN) is preliminarily used for denoising and isolating Intrinsic Mode Functions (IMFs) containing cavitation characteristics from current signals. IMFs revealing pump operation are selected based on the concurrent frequency bands in spectrograms. IMFs are analyzed by Hilbert Transform to select the components that could reveal energy variation, so that the irrelevant components are removed. According to the feature analysis of cavitation, the peak value in marginal spectrum of IMF 6 is extracted to indicate incipient cavitation. And the peak value in marginal spectrum of IMF 7-8 is extracted as the indicator for characterization of cavitation evolution. Results of this research could provide technical assistance for pump cavitation detection.