With the advancement of intelligent industrial equipment and the growing demand for system digitalization, parametric modeling of pump operational states has become increasingly important. This is especially true for large pumps, where real-time monitoring remains a major challenge. This paper proposes a pump characteristic curve prediction method based on transfer learning. By leveraging characteristic curve data from small, easily testable source domains for pre-training, the learned features are transferred as initial conditions for training performance models of other pump types. The test results show that neural network models pre-trained with transfer learning achieve faster prediction speeds and lower error rates. Transfer learning also demonstrates strong adaptability to characteristic curve data from various pump categories. Under varying volumes of target data, prediction accuracy improves significantly. Notably, when data are limited, the transfer learning approach achieves a prediction error of 5.2%, compared to 48.1% for direct deep learning modeling. Moreover, the proposed method effectively reduces prediction range beyond the scope of the original data.

The centrifugal pump with high rotation speed is the key component in the cooling system of an aircraft. Because of the high rotation speed, the impeller inlet is very prone to cavitation. Two impellers with different types of blades (cylindrical and splitter) are designed, and the numerical models of the pumps are built. The authenticity of the numerical models is validated with the corresponding experiments in terms of both the hydraulic and cavitation characteristics. Then, the effects of different coolants and working temperatures on the hydraulic and cavitation performances of the prototype models are studied based on the numerical simulations. The results show that the head and efficiency of the pump for conveying water are higher than those for conveying ethylene glycol (EG) aqueous solution and propylene glycol (PG) aqueous solution (EGaq and PGaq are defined to represent the EG aqueous solution and the PG aqueous solution, respectively). The hydraulic performance in the EGaq is slightly better than that in the PGaq. The cavitation performance of water is far less than that of the EGaq and PGaq under high working temperature. The volume of cavitation in EGaq is smaller than that in PGaq, and the volume of cavitation in the splitter blades is slightly smaller than that in the cylindrical blades. It is suggested that EGaq be used as the first option. The splitter blades can improve the cavitation performance somehow while the improvement by using the splitter blades is very limited at high rotation speeds, and the design of the short blades should be careful in order to obtain a smooth internal flow field.

Traditional centrifugal pumps have limitations such as bearing wear and mechanical seal failures resulting from high-speed operations, which lead to a notable reduction in system efficiency. To overcome these limitations, centrifugal pumps driven by bearingless motors are employed. However, the absence of shaft constraints in such setups causes the impeller rotor to experience offset and tilting, thereby affecting pump performance and potentially causing system malfunctions. Thus, the bearingless direct-drive centrifugal pump (BDDCP) characteristics under extreme operating conditions need to be quantitatively analyzed. This study obtained a suitable turbulence model through experiments and subsequently designed three extreme operating conditions: different flow rates, impeller offset positions, and impeller tilt angles. The velocity field, external characteristics, and radial water excitation force were comprehensively analyzed. The results showed that flow changes greatly impact the BDDCP performance, while tilt angle changes slightly impact the BDDCP performance. These results can serve as a reference for subsequent research on BDDCP under extreme operating conditions.