Showing papers by "Naira Hovakimyan published in 2001"

Adaptive Output Feedback for High-Bandwidth Control of an Unmanned Helicopter

Abstract: We consider adaptive output feedback neuro-control of uncertain nonlinear systems, and in particular its application to high-bandwidt h flight control of unmanned rotorcraft. Given a smooth reference trajectory, the problem is to design a controller that would force the system measurement to track it asymptotically or with bounded errors. The classical approach necessitates building state observers. The state estimates are used both in the controller design and in the adaptation laws. However, finding a good observer for an uncertain nonlinear plant is not an obvious task. We argue that it should be sufficient to build an observer for the output tracking error only. The method is employed in the design of a high-bandwidth attitude command system for an unmanned helicopter.

Adaptive output feedback control of a class of non-linear systems using neural networks

Abstract: This paper presents tools for the design of a neural network based adaptive output feedback controller for a class of partially or completely unknown non-linear multi-input multi-output systems without zero dynamics. Each of the outputs is assumed to have relative degree less or equal to 2. A neural network based adaptive observer is designed to estimate the derivatives of the outputs. Subsequently, the adaptive observer is integrated into a neural network based adaptive controller architecture. Conditions are derived which guarantee the ultimate boundedness of all the errors in the closed loop system. Stability analysis reveals simultaneous learning rules for both the adaptive neural network observer and adaptive neural network controller. The design approach is illustrated using a fourth order two-input two-output example, in which each output has relative degree two.

Adaptive control system having direct output feedback and related apparatuses and methods

Abstract: An adaptive control system (ACS) uses direct output feedback to control a plant. The ACS uses direct adaptive output feedback control developed for highly uncertain nonlinear systems, that does not rely on state estimation. The approach is also applicable to systems of unknown, but bounded dimension, whose output has known, but otherwise arbitrary relative degree. This includes systems with both parameter uncertainty and unmodeled dynamics. The result is achieved by extending the universal function approximation property of linearly parameterized neural networks to model unknown system dynamics from input/output data. The network weight adaptation rule is derived from Lyapunov stability analysis, and guarantees that the adapted weight errors and the tracking error are bounded.

Dynamic neural networks for output feedback control

Abstract: A dynamic neural network is designed to estimate velocities from displacement measurements for a nonlinear system. A linear-in-parameters NN is used for state reconstruction. Conditions are provided under which the estimation error is guaranteed to be ultimately bounded. Subsequently, this observer is integrated into an adaptive controller architecture. The controller is based on model inversion and is augmented with a second learning-while-controlling neural network. Conditions are derived which guarantee ultimate boundedness of all the errors in the combined observer–controller feedback system. Open- and closed-loop simulations for a Van Der Pol oscillator are used to illustrate the results. Copyright © 2001 John Wiley & Sons, Ltd.

A novel observer based adaptive output feedback approach for control of uncertain systems

Abstract: We consider the adaptive output feedback control of nonlinear systems. Given a smooth reference trajectory, the problem is to design a controller that forces the system measurement to track it with bounded errors. The classical approach requires building a state observer. However, finding a good observer for a highly nonlinear and uncertain plant is not an obvious task. We argue that it should be sufficient to build an observer for the output tracking error. The uniform ultimate boundedness of error signals is shown through a Lyapunov stability analysis. Simulations of a nonlinear second order system illustrate the theoretical results.

Decentralized control of large-scale systems using single hidden layer neural networks

Abstract: A decentralized adaptive control design procedure for large-scale uncertain systems is developed using Single Hidden Layer neural networks. The subsystems are assumed to be feedback linearizable and non-affine in the control, and their interconnections bounded linearly by the tracking error norms. Single Hidden Layer neural networks are introduced to approximate the feedback linearization error signal online from available measurements. A robust adaptive signal is required in the analysis to shield the feedback linearizing control law from the interconnection effects. The tracking errors are shown to be uniformly ultimately bounded, and all other signals uniformly bounded. The proposed adaptive algorithm is implemented in simulation to stabilize an interconnected double inverted pendulum.

An SPR approach for adaptive output feedback control with neural networks

Abstract: A direct adaptive output feedback control design procedure is developed for highly uncertain nonlinear systems, that does not rely on state estimation. The approach is also applicable to systems of unknown, but bounded dimension. This includes systems with both parametric uncertainties and unmodelled dynamics. This result is achieved by extending the universal function approximation property of linearly parameterized neural networks to model unknown system dynamics from input/output data. The network weight adaptation rule is derived from Lyapunov stability analysis, that guarantees boundedness of the NN weights and the system tracking errors. Numerical simulations of an output feedback controlled Van der Pol oscillator, coupled with a linear oscillator, are used to illustrate the practical potential of the theoretical results.

Certainty Equivalence Principle and Minimax Team Problems

Abstract: We consider a team problem, with two decision makers for simplicity, where the uncertainties are dealt with in a minimax fashion rather than in a stochastic framework. We do not assume that the players exchange information at any time. Thus, new ideas are necessary to investigate that situation. In contrast with the classical literature, we do not use necessary conditions, but investigate to what extent ideas from the (nonlinear) minimax certainty equivalence theory allow one to conclude here. We are led to the introduction of a “partial-team” problem, where one of the decision makers has perfect state information. We then investigate the full-team problem, but the main result concerning it is shown still to be rather weak. We nevertheless apply it to the linear quadratic case, where it yields an original result.