Neural Network-Based Analysis on the Unusual Peak of Cavitation Performance of a Mixed Flow Pipeline Pump

Cavitation will inevitably occur in the process of the operation of bladed pumps, which is disadvantageous to the pumps. The conventional cavitation research method combines experiment and numerical simulation, but these methods cannot fully reflect the dynamic process of cavitation. Therefore, neural network is applied. In this study, a mixed flow pipeline pump is taken as the research object, and the unusual peak phenomenon of the critical cavitation under small flowrate conditions is analyzed by the combination of experiment, numerical simulation and neural network. The research shows that the unusual peak of critical cavitation mainly occurs under the flowrate range of 0.081-0.108. The application of artificial intelligence can intuitively and accurately reflect the flowrate change of pump and the influence from inception to development of cavitation on pump performance. The appearance of cavitation peak is mainly due to the large fluctuation range of head with the change of flowrate and cavitation degree.

Automated summary

In this study, the phenomenon of unusual peak of critical cavitation under small flowrate conditions is analyzed by the combination of experiment, numerical simulation and neural network. It is found that the unusual peak of critical cavitation mainly occurs under the flowrate range of 0.081-0.108. The application of artificial intelligence can intuitively and accurately reflect the flowrate change of pump and the influence from inception to development of cavitation on pump performance, providing new ideas for improving pump performance and reducing cavitation.