Neurodynamic Programming and Tracking Control Scheme of Constrained-Input Systems Via a Novel Event-Triggered PI Algorithm

Kun Zhang,Huaguang Zhang,Xiaodong Liang,Zhiliang Wang

DOI: https://doi.org/10.1016/j.asoc.2019.105629

IF: 8.7

2019-01-01

Applied Soft Computing

Abstract:In this paper, a new event-triggered approach for neurodynamic programming and optimal tracking control problem of constrained-input systems is proposed, where the desired trajectory can be generated as a large class of useful command trajectories. Firstly, the complex tracking problem is converted to a stabilizing control optimization problem by reconstructing a novel augmented tracking system with discounted performance index. Secondly, instead of the conventional time-driven control, an event-triggered policy iteration (PI) algorithm is designed to drive the system dynamic to track the reference trajectory, which requires less computation and fewer transmissions during the solving. Thirdly, the novel tracking control can be bounded as desired, which overcomes the unconstrained steady-state control from the general adaptive dynamic programming based tracking solution. Moreover, only critic neural network is used in the implementation of iteration learning, which simplifies the actor-critic architecture and reduces computational load. And by the means of Lyapunov method, the ultimately boundedness of the tracking system under the event-triggered PI algorithm is proved. Finally, the developed approach is applied to track a sinusoidal waveform and a periodic rectangular step signal in the simulations, where the effectiveness is also demonstrated.