Neural Network-based Output Feedback Research Articles

Neural network based output feedback control for DC motors with asymptotic stability

Measurement noise and unknown dynamics including friction, parametric uncertainties, external disturbances and unmodeled dynamics, broadly exist in any electromechanical system and worsen their control performance. To address this issue, this paper develops an output feedback control scheme with neural network (NN) based unknown dynamics compensation for DC motor systems. First, to avert using the noise-polluted velocity signal, a state observer with an online adapted gain is adopted, which attenuates the impact of measurement noise on tracking accuracy and reduces the conservatism of observer gain selection. Second, one NN with low computation and ease to design, is employed for unknown dynamics compensation, which is beneficial for practical applications. Then a composite control law is constructed to achieve high-accuracy tracking performance. The stability analysis reveals the tracking error can asymptotically converge to zero while facing time-variant unknown dynamics. Results of comparative experiments validate the superiority of the presented method.

Neuro-adaptive output feedback control of the continuous polymerization reactor subjected to parametric uncertainties and external disturbances

This paper proposes an adaptive neural network based output feedback backstepping fast terminal sliding mode control (NN-BFTSMC) for continuous polymerization reactor with external disturbances and parametric uncertainties. Firstly, neural networks (NN) are employed to approximate the uncertain nonlinear functions. Next, the average molecular weight and the reactor temperature tracking controllers are designed based on the finite-time NN-BFTSMC. In addition, the NN-BFSMC effectively estimates and compensates the upper bounds of the external disturbances. Moreover, the chattering effect is eliminated without losing the robustness accuracy. The finite-time stability of the closed-loop system is proved by Lyapunov theory. At last, numerical simulations and comparative studies are introduced to illustrate the superior performance of the proposed method.

Bounding set calculation for neural network-based output feedback adaptive control systems

Evaluating the bounding set of dynamic systems subject to direct neural-adaptive control is a critical issue in applications where the control system must undergo a rigorous verification process in order to comply with certification standards. In this paper, the boundedness problem is addressed for a comprehensive class of uncertain dynamic systems. Several common but unnecessary approximations that are typically performed to simplify the Lyapunov analysis have been avoided in this effort. This leads to a more accurate and general formulation of the bounding set for the overall closed loop system. The conditions under which boundedness can be guaranteed are carefully analyzed; additionally, the interactions between the control design parameters, the ‘Strictly Positive Realness’ condition, and the shape and dimensions of the bounding set are discussed. Finally, an example is presented in which the bounding set is calculated for the neuro-adaptive control of an F/A-18 aircraft, along with a numerical study to evaluate the effect of several design parameters.