Nonlinear Adaptive Control Using Backpropagating Neural Networks

Abstract

The objective of this research is to develop a nonlinear regulator for an adaptive control system using backpropagating neural networks (BNN's) in conjunction with a linear quadratic regulator (LQR). The basic concepts of adaptive control and the structure of neural networks are discussed. These concepts are integrated and the nonlinear regulator is derived. Simulation is conducted on a representative nonlinear system with both the LQR and the nonlinear regulator. Training of the regulator and its performance under varying BNN parameter values are examined. The simulation results show that the nonlinear regulator with BNN's exhibits superior performance compared to the LQR when the nonlinearities are large. The optimization of regulator performance with regard to BNN parameter values is discussed. Further research is required in order to determine the general applicability of this regulator and to develop more specific guidelines for BNN parameters.

Open PDF

Document Details

Document Type
Technical Report
Publication Date
Jun 01, 1992
Accession Number
ADA256828

Entities

People

  • Kurt W. Menke

Organizations

  • Naval Postgraduate School

Tags

Communities of Interest

  • Energy and Power Technologies

DTIC Thesaurus Topics

  • Adaptive Control Systems
  • Algorithms
  • Computers
  • Control Systems
  • Control Systems Engineering
  • Differential Equations
  • Engineering
  • Equations
  • Neural Networks
  • Nonlinear Systems
  • Riccati Equation
  • Schools
  • Simulations
  • Training
  • Transfer Functions
  • United States
  • United States Naval Academy

Readers

  • Calculus or Mathematical Analysis
  • Molecular and Cellular Biology
  • Robotics and Automation.

Technology Areas

  • AI & ML
  • AI & ML - Bayesian Inference
  • AI & ML - DoD AI Strategy
  • AI & ML - Machine Learning Algorithms
  • AI & ML - Neural Networks