SNAC convergence and use in adaptive autopilot design

  • Authors:

  • Songjie Chen;Yang Yang;S. N. Balakrishnan;Nhan T. Nguyen;K. Krishnakumar

  • Affiliations:

  • Department of Mechanical and Aerospace Engineering, Missouri University of Science and Technology;Department of Mechanical and Aerospace Engineering, Missouri University of Science and Technology;Department of Mechanical and Aerospace Engineering, Missouri University of Science and Technology;NASA Ames Research Center, Moffett Field, CA;NASA Ames Research Center, Moffett Field, CA

  • Venue:
  • IJCNN'09 Proceedings of the 2009 international joint conference on Neural Networks
  • Year:
  • 2009

Quantified Score

Hi-index 0.00

Visualization

Abstract

In this paper, approximate dynamic programming (ADP)based design tools are developed for adaptive control of aircraft control under nominal and damaged conditions. Nominal control of the system is computed with a Single Network Adaptive Critic(SNAC) derived through principles of ADP. Convergence of SNAC training is shown by reducing it to solving a set of nonlinear algebraic equations in weights. Unlike many adaptive control approaches, we develop approximate optimal control expressions to handle uncertainties. Uncertainties are calculated with an online neural network with guaranteed convergence. Longitudinal dynamics of an aircraft is used to illustrate the working of the developed algorithms.