Loading...

Neural network prediction of a liquid-liquid coaxial swirl injector performance map

Ghorbanian, K ; Sharif University of Technology | 2005

255 Viewed
  1. Type of Document: Article
  2. DOI: 10.2514/6.2005-1430
  3. Publisher: American Institute of Aeronautics and Astronautics Inc , 2005
  4. Abstract:
  5. A general regression neural network technique is applied for design optimization of liquid-liquid coaxial swirl injectors. Phase Doppler Anemometry measurements for velocity and SMD distributions for various Reynolds numbers are used to train the neural network to generate the injector's performance map. Excellent agreement between the predicted values and the measurements is obtained. It is observed that by reducing the number of randomly selected training samples to about one third of the number of prediction points, one may reconstruct the velocity field in the extrapolation regime with an accuracy of 93%. Copyright © 2005 by the American Institute of Aeronautics and Astronautics, Inc. All rights reserved
  6. Keywords:
  7. Anemometers ; Neural networks ; Performance ; Reynolds number ; Coaxial swirl injector ; Performance map ; Velocity fields ; Swirling flow
  8. Source: 43rd AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, 10 January 2005 through 13 January 2005 ; 2005 , Pages 11401-11408
  9. URL: https://arc.aiaa.org/doi/10.2514/6.2005-1430