TU BRAUNSCHWEIG

Flight Systems Technology

The research activities in aircraft system technology are focused on the development, analysis of robustness, and acceptance evaluation of additional automated flight control functions for commercial aircraft, and their advantages in all-weather operation within the congested airspace near cities. In addition, new approaches in pilot support are being evaluated through testing in a flight simulator and their impact on flight safety researched. As part of the research program, an automated decision-making system for aborting landings will be examined. The evaluation of additional assistance systems, also in terms of future flight scenarios given increasing on-board and ground automation, requires the flight system and its dynamic interaction with humans to be looked at holistically.

At present, the A320-based aircraft simulator is being used for preliminary investigations and the corresponding avionics systems are being implemented. At a later time, the motion-based simulator currently under construction will be put to use.

In addition, a systematic design of hybrid, adaptive flight control strategies will bring together the advantages of model-based controllers with the learning ability seen in artificial neural networks. In addition to improving control behavior in normal operation, the ability of neural networks to adapt to changing conditions will support pilots in high-grade non-linear flight situations (e.g. during reconfiguration of degraded systems). This includes acquiring qualified basic knowledge, training reactions under extreme conditions, as well as developing a suitable sensor concept.

At present, the network structure for the neural network has been defined and the training process successfully implemented. The selected network structure demonstrated fast convergence and the stability of the control was successfully verified.

For subsequent investigations, a pilot model for landing activities based on neural networks will be created, with which the work divided between the automatic control and the human pilot can be evaluated. Using this “virtual pilot”, the decision-making process can be modeled for the tasks of approach and landing.


  last changed 21.12.2011
TU_Icon_E_Mail_1_17x17_RGB pagetop