ABSTRACT |
Methods of testing aircraft for departures range from simple, single-parameter criteria to complex, in-flight departure resistance maneuvers. These methods are useful for predicting departure characteristics, but single-parameter methods may be limited in accuracy because of simplifying assumptions made in their derivation. Also, in-flight or simulation testing of departure resistance maneuvers can be limited by the small number of conditions tested. These limitations increase at high angles of attack where the dynamics of the aircraft are more complex. This paper presents a method for using genetic algorithms to augment traditional evaluation criteria. Quasi-random control inputs are generated by a genetic algorithm for a high fidelity X-31 simulation. Each input is evaluated to determine if it causes a departure. The result of the genetic-algorithm-based search is a population, or set, of control input combinations that lead to uncontrolled flight conditions in the simulation. Recognizing possible differences and simplifications between simulation models and the real aircraft, the results show that the method used is effective for finding possible departures caused by inertial coupling and aerodynamic asymmetries. Simulation data are used to show the results of the genetic algorithm search.
|