Aeroengine Direct Thrust Inverse Control Based On Neural Networks

The main functions of aero-engines were to provide thrust for aircrafts. So the problem how to control thrust effectively was most interesting thing to designers. Because engine thrust was un-measurable directly when the aircraft was flying,in conventional aero-engine control systems some measurable parameters, such as rotor speed and pressure ratio, were used as controlled variables. The controlled variables could be used to calculate the real interested parameter—engine thrust. Due to the un-accuracy of the calculated thrust, large design margins have to be built to protect aero-engines. That is to say that we have to"sacrifice"the partial performance to ensure aero-engines running normally.In this paper, the investigations about direct thrust control of aero-engines have been made for exploiting the potential performance of aero-engines sufficiently. For the realization of direct thrust control of aero-engines, firstly, aero-engines thrust estimator was studied and designed. The author presented a new method named Adaptive Genetic Neural Network Algorithm (AGNNA). The method was applied to design the thrust estimator successfully based on a turbo-fan aero-engine. Secondly, the augmented LQR controller which can eliminate steady state error of controlled variables was set up. So the direct thrust control system was completed based on the augmented LQR controller and the thrust estimator. Lastly, because of quickly development of neural network inverse control, especially in the domain of nonlinear control, the author attempted to apply it to aero-engine control. A new aero-engine controller based on neural network inverse control was investigated and designed in this paper. So another direct thrust control system of aero-engines was brought to success in the basis of the neural network inverse controller and the thrust estimator by digital simulation and validation.