Share Email Print

Proceedings Paper

Numerical model of adaptive optical system controlled by a feedforward neural network
Author(s): Gleb V. Vdovin
Format Member Price Non-Member Price
PDF $17.00 $21.00

Paper Abstract

An adaptive optical system (AOS) with a feedback loop closed via feedforward neural network (NN) is considered. The vector of the wavefront corrector control signals is computed by the network from two vectors of the intensity moments measured in two near-field planes by two matrix photo-detectors. The NN is trained with back-propagation algorithm to predict the vector of AM signals from the measured intensity vectors. During training phase the network forms a control algorithm for a given configuration of the optical system, taking into account misalignments and nonlinearities of the hardware used. A numerical model of a multichannel AOS controlled by a multilayer NN has been built, trained, and run for different low-order input aberrations. The neural control permits a direct conversion of the intensity distribution measured in the near field into control signals of the wavefront corrector. High efficiency of control has been demonstrated for a model of a 16-channel adaptive optical system for arbitrary input aberrations having limited spatial spectrum.

Paper Details

Date Published: 25 August 1995
PDF: 8 pages
Proc. SPIE 2534, Adaptive Optical Systems and Applications, (25 August 1995); doi: 10.1117/12.217752
Show Author Affiliations
Gleb V. Vdovin, Delft Univ. of Technology (Netherlands)

Published in SPIE Proceedings Vol. 2534:
Adaptive Optical Systems and Applications
Robert K. Tyson; Robert Q. Fugate, Editor(s)

© SPIE. Terms of Use
Back to Top