A two-neural network approach to solving optimal control problems is described in this study. This approach called the adaptive critic method consists of two neural networks: one is called the supervisor or critic, and the other is called an action network or controller. The inputs to both these networks are the current states of the system to be controlled. Each network is trained through an output of the other network and the conditions for optimal control. When their outputs are mutually consistent, the controller network output is optimal. The optimality is limited to the underlying model. Hence, we develop a Lyapunov based theory for robust stability of these controllers when there is input uncertainty. We illustrate this approach through a longitudinal autopilot of a nonlinear missile problem.
S. N. Balakrishnan and Z. Huang, "Robust Adaptive Critic Based Neurocontrollers for Systems with Input Uncertainties," Proceedings of the IEEE-INNS-ENNS International Joint Conference on Neural Networks, 2000, Institute of Electrical and Electronics Engineers (IEEE), Jan 2000.
The definitive version is available at http://dx.doi.org/10.1109/IJCNN.2000.861282
IEEE-INNS-ENNS International Joint Conference on Neural Networks, 2000
Mechanical and Aerospace Engineering
Keywords and Phrases
Lyapunov Method; Lyapunov Methods; Adaptive Critic Method; Missile Guidance; Neural Network; Neurocontrollers; Optimal Control; Robust Control; Stability; Uncertain Systems
Article - Conference proceedings
© 2000 Institute of Electrical and Electronics Engineers (IEEE), All rights reserved.