Based on a test apparatus of the on-board hollow fiber membrane separation performance

the separation performance was experimentally investigated with variation of the bleed air pressure

bleed air temperature

flight altitude and other factors; the experimental data were applied into the artificial neural network prediction technology as the training and validation samples

so as to predict the performance of this membrane. The results show that: (1)the established mathematical model can effectively predict the separation performance of hollow fiber membrane; (2)the volume fraction of nitrogen-enriched air is inversely proportional to non-dimensional flow rate; so when the volume fraction of nitrogen-enriched air increases

the flow rate and efficiency of nitrogen-enriched air decrease; (3)under a given volume fraction of nitrogen-enriched air

non-dimensional flow rate of nitrogen-enriched air increases with bleed air temperature and bleed air pressure increasing; the efficiency of nitrogen-enriched air increases with the bleed air pressure increasing

but decreases with bleed air temperature increasing; (4) when the flight altitude increases

the non-dimensional flow rate and efficiency of nitrogen-enriched air increase

and the influence of volume fraction of nitrogen-enriched air decreases with the growing flight altitude.