Intelligent adaptive nonlinear flight control for a high performance aircraft with neural networks.

Area of Science:

• Aerospace Engineering
• Control Systems
• Artificial Intelligence

Background:

• Traditional flight control systems face challenges with system uncertainties, dynamic flight conditions, and component failures.
• High-performance aircraft require advanced control systems capable of real-time adaptation and robustness.

Purpose of the Study:

• To develop a neural network (NN)-based adaptive flight control system for a nonlinear F-16 aircraft model.
• To enhance flight control by compensating for system uncertainties, adapting to changing flight conditions, and accommodating system failures.

Main Methods:

• Developed a NN-based adaptive identification model for aircraft angular rates using on-line training and the Levenberg-Marquardt optimization method.
• Created a NN-based adaptive PID control scheme comprising an emulator NN, an estimator NN, and a discrete-time PID controller.
• Implemented an on-line training procedure using a first-in-first-out stack for input-output data and adapted the Levenberg-Marquardt optimization for NN training.

Main Results:

• The NN-based adaptive identification model accurately captured the dynamic behavior of the nonlinear F-16 model.
• The NN-based adaptive PID control system demonstrated effective control of pitch, roll, and yaw rates.
• The developed control system exhibited learning, adaptation, and fault-tolerant capabilities, outperforming traditional methods by avoiding extensive parameter storage.

Conclusions:

• The proposed NN-based adaptive flight control system offers a robust and intelligent solution for high-performance aircraft.
• The system's ability to learn, adapt, and tolerate failures significantly improves flight safety and performance.
• This approach provides a more efficient alternative to conventional flight control systems by reducing the need for extensive parameter management.