To diagnose common failures in vertical Essential Service Water Pumps (SEC), a method combining the wavelet packet transform (WPT) and the support vector machine (SVM) was adopted. This allowed us to construct a diagnostic model capable of classifying multiple states, including the six types of faults and normal conditions in SEC pumps. The diagnostic model utilized the wavelet packet coefficients to capture sub-bands with a higher energy share and reconstruct the signals. The model inputs the 12 frequency features into the support vector machine to analyze the vibration signals gathered from the SEC pump benchmark. The study illustrates that the proposed method can accurately differentiate between various fault conditions when compared to the WPT method, combined with the artificial neural network (ANN) approach. It attains a superior overall precision of up to 94%, and it displays excellent generalization and strong adaptability.

This paper proposes a data-driven prediction scheme for the remaining life of centrifugal pump bearings based on the KPCA–LSTM network. A centrifugal pump bearing fault experiment bench is built to collect data, and the performance of time domain, frequency domain, and time-frequency domain characteristics under different working conditions is analyzed. Time domain characteristics, frequency domain characteristics, wavelet packet decomposition energy characteristics, and CEEMDAN energy features are found to be able to capture fault information under different working conditions. Therefore, 43 sensitive features are determined from the time domain, frequency domain, and time-frequency domain. Through the analysis of XJTU-SY bearing life cycle data and based on the weighted scores of monotonicity, robustness, and trend indicators, twelve outstanding characteristics of the bearing in the whole life cycle are selected, and a one-dimensional feature quantity that can characterize the life-degradation process of the centrifugal pump bearing is constructed after KPCA dimension reduction processing. The LSTM network, sensitive to time series, is selected to predict and analyze the constructed one-dimensional feature trend, and the prediction effects of the BP network and the CNN network are compared. The results show that this method has advantages in prediction accuracy and model training time.

The precise identification of faults in centrifugal pumps is crucial for ensuring their safe and stable operation, given their significance as vital industrial equipment. This article aims to rigorously examine and analyze the flow characteristics of centrifugal pumps under two specific conditions: normal blade operation and a single blade breakage fault. Through systematic comparison and in-depth study, this article sheds light on distinguishing flow patterns exhibited by these pumps under both normal and fault scenarios. Utilizing validated numerical simulation methods, a thorough analysis is conducted to explore the flow condition and energy characteristics of the impeller channel following the breakage of a single blade. Additionally, the article investigates changes in the pressure pulsation characteristics of the pump volute as a result. The numerical simulation results reveal that the head of the centrifugal pump decreases at all flow points when a single blade breaks. However, there is no significant change in efficiency at small flow points. As the flow rate exceeds 0.9Q_d, efficiency experiences a substantial decrease. Furthermore, the efficiency decline becomes even more pronounced as the flow rate continues to increase. At 1.5Q_d, efficiency plummets by 14.33%. The flow pattern undergoes significant changes as well. The breakage of the blade induces noteworthy secondary flow in adjacent impeller channels, resulting in heightened turbulence dissipation. Additionally, it was observed that blade fracture causes alterations in the main frequency of pressure pulsation within the volute. This is characterized by an increase in shaft passing frequency and a decrease in blade passing frequency. Notably, near the tongue monitoring point, the shaft frequency escalates by over 20 times.