Structural Optimization Design and Experimental Study of Centrifugal Blood Pump Impeller.
TL;DR
The NN-GA demonstrates superior global optimization capabilities to the orthogonal experiment and guides pump design in energy, chemical, and semiconductor industries, providing new insights for multi-parameter coupled centrifugal blood pump optimization.
Abstract
Objective
As a core component of mechanical circulatory support devices, a centrifugal blood pump with excessively low hydraulic efficiency can lead to poor blood compatibility, low efficiency, excessive heat generation, insufficient blood supply, and pressure fluctuations. These issues adversely affect patient prognosis. To address these bottlenecks, this study aims to perform structural optimization of the centrifugal blood pump with the goal of improving hydraulic efficiency and exploring suitable optimization methods.
Methods
This study takes the UJN-1 magnetically suspended centrifugal blood pump as the research object, with hydraulic efficiency simulated by computational fluid dynamics (CFD) as the target, and compares the results of optimizing four impeller parameters (blade inlet angle, outlet angle, number of blades, blade height) using orthogonal experiments and neural network-genetic algorithm (NN-GA).
Results
Genetic algorithm (GA) optimization yielded an impeller with 20.48° inlet angle, 17.22° outlet angle, 6 blades, and 4.55 mm height. Its hydraulic efficiency increased by 2.38%, outlet pressure fluctuation decreased significantly, and it outperformed the orthogonal pump by 1.57%. Hydraulic experiments validated the results, with deviations within approximately ±1% of CFD.
Conclusion
The NN-GA demonstrates superior global optimization capabilities to the orthogonal experiment.
Significance
This study provides new insights for multi-parameter coupled centrifugal blood pump optimization and guides pump design in energy, chemical, and semiconductor industries.