Title :
A new class of modular adaptive controllers, Part II: Neural network extension for non-LP systems
Author :
Patre, P.M. ; Dupree, K. ; MacKunis, W. ; Dixon, W.E.
Author_Institution :
Dept. of Mech. & Aerosp. Eng., Univ. of Florida, Gainesville, FL
Abstract :
The development in this (Part II) paper augments the result developed in Part I by considering uncertain dynamic systems that are not necessarily linear-in-the-parameters (LP), and have additive non-LP bounded disturbances. For non-LP uncertainties, a model-based adaptive feedforward formulation cannot be used. Therefore, in this paper, a multilayer neural network (NN) structure is used as a feedforward element (that learns and compensates for the non-LP dynamics) in conjunction with the Robust Integral of the Sign of the Error (RISE) feedback term. Similar to the result in Part I, a NN- based controller is developed in this paper with modularity in NN weight tuning laws and control law. Specifically, the results in this paper allow the NN weight tuning laws to be determined from a developed generic update law (rather than be restricted to a gradient update law).
Keywords :
adaptive control; closed loop systems; control nonlinearities; control system synthesis; multilayer perceptrons; neurocontrollers; nonlinear control systems; uncertain systems; NN weight control law; NN weight tuning laws; NN-based controller; closed-loop error system; feedforward element; linear-in-the-parameters; modular adaptive controllers; multilayer neural network; nonLP systems; nonlinear multilayer NN structure; uncertain dynamic systems; Adaptive control; Aerodynamics; Control systems; Error correction; Feedforward neural networks; Multi-layer neural network; Neural networks; Programmable control; Stability analysis; Uncertainty;
Conference_Titel :
American Control Conference, 2008
Conference_Location :
Seattle, WA
Print_ISBN :
978-1-4244-2078-0
Electronic_ISBN :
0743-1619
DOI :
10.1109/ACC.2008.4586658
