Adaptive NN Observer-Based Synthesize Strategy for Connected Nonlinear System

Abstract

In this paper, the issues of tracking and synchronization for connected nonlinear servo system is considered. An adaptive neural network (NN) observer-based synthesize strategy is proposed. Firstly, a mathematical model of connected nonlinear isomorphism multi-motor servo system with disturbance is established, where the motors are connected through wired or wireless networks. Then, an adaptive disturbance observer based on sliding-mode is proposed, and the finite time convergence of observation errors has been proven. Thirdly, an adaptive NN tracking strategy is designed, and we have proved the tracking error is bounded and the full-state asymmetric constraints are satisfied. Furthermore, a synchronous control technique on sliding-mode is presented, and both the boundedness of synchronous error and reachability of sliding surface are verified. Finally, a numerical simulation and a semi-physical simulation are carried out to illustrate the validity of proposed methods. Note to Practitioners—This paper was motivated by the tracking and synchronization of connected nonlinear servo system, which encounters with the adverse effect of disturbance. We present a novel adaptive NN observer-based synthesize strategy to achieve the desired control performance. First of all, we established the dynamic of connected nonlinear isomorphism multi-motor servo system under disturbance. In order to obtain prior knowledge of disturbances, we design an adaptive sliding-mode disturbance observer and the observation errors can converge to a small domain within finite time. To ensure that all motors can track the desired trajectory without violating the full state constraints, we propose an adaptive NN tracking scheme based on coordinate transformation and the boundedness of the tracking error is demonstrated. Additionally, to achieve synchronization of all motors, we present a sliding-mode synchronous control approach, and the reachability of sliding surface and the boundedness of synchronous error are proved, simultaneously. Both simulation and experiment indicate that designed algorithms are effective and feasible.