To investigate the flow field characteristics of a low specific speed axial flow pump under different flow conditions ( Q/ Q _des=0.9, 1.0, 1.1), the particle image velocimetry (PIV) technique is employed to measure the axial cross-sections of the pump at four locations: upstream of the impeller, inside the impeller, inside the guide vane, and downstream of the guide vane. And the pressure transducer is used to obtain the pressure distribution of the wall in the flow field of the pump. The results indicate that the velocity distribution upstream of the impeller shows consistent trends with axial acceleration along the axial direction under all three working conditions. A large velocity gradient is observed inside the impeller at different phases, displaying obvious acceleration near the hub along with concentrated high-speed regions. The velocity distribution inside the guide vane is similar, with more uniform high-speed regions and more stabilized flow under the design flow rate. The flow velocity downstream of the guide vane gradually increases in the mainstream direction, slowly accelerating at low flow rates and rapidly approaching the velocity in the mainstream region at high flow rates. As the flow rate increases, the pressure values at the monitoring points in the pump flow channel decrease. The main frequency of the pressure fluctuation is the blade passing frequency. And the pressure transition between the impeller and the guide vane gap is relatively smooth under the design flow rate with reduced flow losses. The experimental results can provide references for the design and performance optimization of the axial flow pump.

An ensemble deep transfer learning method for fault diagnosis based on soft voting is proposed for diagnosing waterjet pump faults under variable working conditions. The source domain and few target domain data samples are normalized after FFT transformation and then fed into three deep transfer learning diagnosis models for training: the CORAL based deep transfer metric learning model, the MMD based deep transfer metric learning model, and the transfer component-based deep belief network. The target domain test samples are diagnosed and analyzed based on this approach. An ensemble deep transfer diagnosis model is subsequently established by combing the soft voting method to obtain the final diagnosis results. Through the diagnosis of three different types of faults in waterjet pumps under variable working conditions, the results show that the proposed ensemble deep transfer diagnosis model not only effectively addresses the high-precision fault diagnosis of waterjet pumps under variable working conditions, but also has better diagnostic accuracy than the single deep transfer fault diagnosis model.