.png){kind=icon}
.png){kind=icon}
Return
Neural-network-based control via adaptive dynamic programming under dual description coding
.png){kind=icon}
PRE
AI
DOI：10.1016/j.nahs.2026.101706.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
This paper investigates the adaptive control problem based on neural networks (NN) for a class of multi-input multi-output nonlinear discrete-time systems. Since the system dynamic model is unknown and only input–output data are available, the unknown nonlinear dynamic model is first transformed into an equivalent dynamic linear model by using the pseudo-partial derivative (PPD) technique. Then, to enhance the reliability of signal transmission, a dual-description coding scheme is implemented, encoding the system outputs into two equally important descriptions. An offline value-iteration-based adaptive dynamic programming (ADP) algorithm is subsequently developed to address the optimal control problem based on the coding–decoding scheme. The convergence of the proposed method is rigorously analyzed using mathematical induction. Furthermore, within the ADP framework, an actor–critic NN method is proposed to approximate both the control law and the performance index function. To guarantee the boundedness of the real tracking errors and the estimate errors of critic and actor NN weights, some sufficient conditions are derived. Finally, numerical simulations are performed to validate the effectiveness of the proposed control strategy.
Keywords:
Neural networks
Adaptive dynamic programming
Dual-description coding
Nonlinear discrete-time systems
Optimal control
Journal
N
IF:
0
Papers:
94
Citations:
0