Cavitation Diagnosis Method for Centrifugal Pumps based on Agglomerative Hierarchical Clustering Algorithm

Abstract

Cavitation in the pump induces vibration and noise, which leads to the degradation of pump performance and damage to the impeller. Therefore, it is of great significance to accurately identify the cavitation state of the centrifugal pump. To determine the cavitation state using the vibration characteristic frequency of the centrifugal pump, it is necessary to accurately identify the characteristic frequency due to the noise's effect. Therefore, a cavitation fault diagnosis method of centrifugal pump based on a cohesive hierarchy algorithm was proposed. Firstly, Singular Value Decomposition (SVD) is used to de-noise the vibration signal. Then the root means square of the vibration signal after de-noising is obtained as the eigenvalue of the vibration signal. Secondly, the hierarchical clustering algorithm is used to classify the vibration eigenvalues and accurately identify the non-cavitation and cavitation states of centrifugal pumps. Finally, through experimental tests, the method can effectively and quickly identify the cavitation state of the centrifugal pump, with an accuracy of 95%. The study provides a new way for rapidly diagnosing centrifugal pump cavitation.