| The object of research is an aircraft model provided by some research institute. The emphasis is laid on the research of self-repairing flight control laws to make fighter land safely. some control laws, include control laws under normal condition, control laws under fault condition and self-repairing control laws under fault condition in the VC++ environment.Firstly, according to the principles of dividing timing marks, the fighter model is divided into two different loops of which one is faster and the other is slower: the faster loop is attitude control loop and the law is model reference adaptive tracking control based on neural network dynamic inversion (NNDI). And, to adaptively regulate the error in the plant inversion which may be caused by modeling uncertainties and disturbances, self-organizing fuzzy cerebella model articulation controller (SOFCMAC) neural network is used. The slower loop is to control flight path. Among them,the longitude path control laws is realized through kinematical characteristic of fighter, and the lateral path control laws is realized by using neural network reorganization and control theory.Secondly, when the rudders of fighter go wrong, in order to improve the security of landing, theories of adaptive feedback output control and adaptive neural network compensator were used to increase the stability of attitude control loop during the process of landing. And later some simulation about various failure happened to fighter during the landing process were given.Thirdly, models of wind field disturbance are used in the fighter model. Through simulation, the result is validated that the self-repairing control laws of safe landing adopt in this thesis can attain the goals of design under the disturbance of wind force of definite level.Finally, some modules and their function in the real-time flight simulation platform were introduced briefly. And the control laws designed in this thesis were run on the platform to validate their effect.
|
PDF Full Text Request