Centrifugal pumps are commonly utilized in thermo-fluidic systems in the industry. Being a rotating machinery, they are prone to vibrations and their premature failure may affect the system predictability and reliability. To avoid their premature breakdown during operation, it is necessary to diagnose the faults in a pump at their initial stage. This study presents the methodology to diagnose fault of a cent rifugal pump using two distinct machine learning techniques, namely, Support vector machine (SVM) and Artificial neural network (ANN). Different statistical features are extracted in the time and the frequency domain of the vibration signal for different working conditions of the pump. Furthermore, to decrease the dimensionality of the obtained features different feature ranking (FR) methods, namely, Chi-square, ReliefF and XGBoost are employed. ANN technique is found to be more efficient in classifying faults in a centrifugal pump as compared to the SVM, and Chi-square and XGBoost ranking techniques are better than ReliefF at sorting more relevant features. The results presented in thus study demonstrate that an ANN based machine learning approach with Chi-square and XGBoost feature ranking techniques can be used effectively for the fault diagnosis of a centrifugal pump.

No abstract

No abstract

Rags, dusts, foreign particles etc. are primary cause of blockage in centrifugal pump and deteriorates the performance. This study elaborates an experimental and data-driven methodology to identify suction, discharge and simultaneous occurrence of both blockages. The discharge pressure signals are acquired and denoised using CEEMD. The fuzzy recurrence plots obtained from denoised signals are attempted to classify using three pre-trained models: Xception, GoogleNet and Inception. None of these models are trained on such images, thus, features are extracted from different pooling layers which include shallow features too. The features extracted from different layers are fed to four shallow learning classifiers: Quadratic SVM, Weighted KNN, Narrow Neural network, and subspace discriminant classifier. The study finds that subspace discriminant achieves highest accuracy of 97.8% when trained using features from second pooling of Xception model. Furthermore, this proposed methodology is implemented at other blockage condition of the pump. The subspace discriminant analysis outperforms the other selected shallow classifier with an accuracy of 93% for the features extracted from the first pooling layer of the Xception model. Therefore, this study demonstrates an efficient method to identify pump blockage using pre-trained and shallow classifiers.

No abstract