Robust Neurocontrollers for Systems with Model Uncertainties: A Helicopter Application

A two-neural network approach to solving nonlinear optimal control problems is described. This approach, called the adaptive critic method, consists of one neural network, called the supervisor or the critic, and a second network, called an action network or a controller. The inputs to both these networks are the current states of the system to be controlled. Targets for each network updates are obtained with outputs of the other network, state propagation equations, and the conditions for optimal control. When their outputs are mutually consistent, the controller network output is optimal. The optimality is, however, limited by the underlying system model. Hence, a Lyapunov theory-based analysis for robust stability of the system under model uncertainties is developed and an extra control is developed. This extra control added with the basic control effort from the adaptive critic method guarantees good system performance and stablity under model uncertainties. This approach is demonstrated through a helicopter problem.

[1]  Robert M. Sanner,et al.  Gaussian Networks for Direct Adaptive Control , 1991, 1991 American Control Conference.

[2]  P.J. Werbos,et al.  An overview of neural networks for control , 1991, IEEE Control Systems.

[3]  Kumpati S. Narendra,et al.  On a new approach to the design of tracking controllers for nonlinear dynamical systems , 1999, Proceedings of the 1999 American Control Conference (Cat. No. 99CH36251).

[4]  Anthony J. Calise,et al.  Analysis of Adaptive Neural Networks for Helicopter Flight Control , 1997 .

[5]  John Preskill.  Making Weirdness Work: Quantum Information and Computation , 2006 .

[6]  Eduardo Sontag,et al.  On characterizations of the input-to-state stability property , 1995 .

[7]  Anthony J. Calise,et al.  Nonlinear adaptive flight control using neural networks , 1998 .

[8]  Kumpati S. Narendra,et al.  Identification and control of dynamical systems using neural networks , 1990, IEEE Trans. Neural Networks.

[9]  Shaheen Ahmad,et al.  Dynamical sliding mode control approach for vertical flight regulation in helicopters , 1994 .

[10]  Eduardo Sontag.  Smooth stabilization implies coprime factorization , 1989, IEEE Transactions on Automatic Control.

[11]  Kumpati S. Narendra,et al.  Identification and control , 1998 .

[12]  S. N. Balakrishnan,et al.  Adaptive-critic based neural networks for aircraft optimal control , 1996 .

[13]  William L. Garrard,et al.  Nonlinear inversion flight control for a supermaneuverable aircraft , 1992 .

[14]  Frank L. Lewis,et al.  Multilayer neural-net robot controller with guaranteed tracking performance , 1996, IEEE Trans. Neural Networks.

[15]  M. B. McFarland,et al.  Robust adaptive control using single-hidden-layer feedforward neural networks , 1999, Proceedings of the 1999 American Control Conference (Cat. No. 99CH36251).

[16]  Richard S. Sutton,et al.  Reinforcement Learning , 1992, Handbook of Machine Learning.

[17]  Arthur E. Bryson,et al.  Applied Optimal Control , 1969 .

[18]  Anthony J. Calise,et al.  Nonlinear flight control using neural networks , 1994 .

[19]  Eric N. Johnson,et al.  FEEDBACK LINEARIZATION WITH NEURAL NETWORK AUGMENTATION APPLIED TO X-33 ATTITUDE CONTROL , 2000 .

[20]  Roger W. Brockett,et al.  Feedback Invariants for Nonlinear Systems , 1978 .

[21]  A. Schaft.  On a state space approach to nonlinear H ∞ control , 1991 .

[22]  A. Schaft.  L/sub 2/-gain analysis of nonlinear systems and nonlinear state-feedback H/sub infinity / control , 1992 .

[23]  L. Hunt,et al.  Global transformations of nonlinear systems , 1983 .

[24]  W. P. Mounfield,et al.  High-gain PI control of an aircraft lateral control system , 1993, Proceedings of IEEE Systems Man and Cybernetics Conference - SMC.

[25]  George Meyer,et al.  Application of nonlinear transformations to automatic flight control , 1984, Autom..