Electromagnetic surface error compensation for reflector antennas using neural network computing

The feasibility of using neural network computing to perform constrained least squares (CLS) surface error compensation has been demonstrated. The major advantage of using the neural-network approach is that, once trained, the large computational overhead associated with the CLS algorithm is overcome and real-time compensation is facilitated. The complex excitations for the surface error compensation were computed using surface data without any field information. Measured field data could, however, also be used to train the network.<>