Complete landing autopilot having control laws based on neural networks and dynamic inversion

This paper discusses the automatic control of airplane in longitudinal and lateral-directional planes, during landing, using two neural network and dynamic inversion concept based adaptive control subsystems, having as components: linear observers, linear dynamic compensators, reference models, and Pseudo Control Hedging blocks in order to eliminate the adapting difficulties of the neural network appearing because of the actuators' nonlinearities. Two subsystems (for longitudinal and lateral-directional planes) will be put together and a new complete landing auto-pilot will result; it is software implemented the results' accuracy and the achievement of the precision standards being briefly analyzed.