ARTIFICIAL NEURAL NETWORKS IN-SPACE STATION OPTIMAL ATTITUDE-CONTROL

Innovative techniques of using ''artificial neural networks'' (ANN) fo r improving the performance of the pitch axis attitude control system of Space Station Freedom using control moment gyros (CMGs) are investi gated. The first technique uses a feed-forward ANN with multi-layer pe rceptrons to obtain an on-line controller which improves the performan ce of the control system via a model following approach. The second te chnique uses a single layer feed-forward ANN with a modified back prop agation scheme to estimate the internal plant variations and the exter nal disturbances separately. These estimates are then used to solve tw o differential Riccati equations to obtain time varying gains which im prove the control system performance in successive orbits.