Doubly curved aperture antenna shape prediction with the use of network-based predictors

• Published in: Microwave and optical technology letters Vol. 33; no. 3; pp. 156 - 163
• Main Authors: Punhani, Amitesh, Washington, Gregory, Theunissen, Wilhelmus H.
• Format: Journal Article
• Language: English
• Published: New York Wiley Subscription Services, Inc., A Wiley Company 05-05-2002
• Subjects: active subreflectors; neural networks; reconfigurable antennas; recursive least squares
• ISSN: 0895-2477; 1098-2760
• DOI: 10.1002/mop.10263

The main objective of this work is to predict the shape of an antenna subreflector that produces a desired radiation far‐field pattern by utilizing artificial intelligence and other methodologies. In this study the size of the radiation beam is kept constant while it is steered throughout the domain. Three different methodologies or constructs are used to develop this model: Neural networks, batch least squares, and recursive least squares. The accuracy of a method is measured by the sum‐squared error of the training examples. During training the variables inside of the constructs are varied, so that the predicted aperture shape matches the actual shape. The networks predicted the antenna reflector shape at an average accuracy of over 97%, the maximum being 99.78%. © 2002 Wiley Periodicals, Inc. Microwave Opt Technol Lett 33: 156–163, 2002; Published online in Wiley InterScience (www.interscience.wiley.com). DOI 10.1002/mop.10263