Cognitive Control Using Adaptive RBF Neural Networks and Reinforcement Learning for Networked Control System Subject to Time-Varying Delay and Packet Losses

• Published in: Arabian journal for science and engineering (2011) Vol. 46; no. 10; pp. 10245 - 10259
• Main Authors: Wang, Shuti, Yin, Xunhe, Li, Peng, Zhang, Yanxin, Wang, Xin, Tong, Shujie
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.10.2021; Springer Nature B.V
• Subjects: Adaptive control; Control methods; Control systems design; Controllers; Engineering; Entropy (Information theory); Humanities and Social Sciences; Machine learning; multidisciplinary; Network control; Neural networks; Proportional integral derivative; Research Article-Systems Engineering; Robust control; Science; Time lag; Time varying control; Networked control system; Adaptive RBF neural networks; Q-learning; PID; Cognitive control
• ISSN: 2193-567X; 1319-8025; 2191-4281
• DOI: 10.1007/s13369-021-05752-y

This paper proposes a novel cognitive control strategy for overcoming the impacts of time-varying delay and data packet losses in networked control system. The Bernoulli distribution is used to characterize the packet losses and time-varying delay. Then, the information entropy is employed for computing the corresponding uncertainties and describe the information gap in the cognitive control. With Q-learning, PID and adaptive RBF neural networks, an improved cognitive control is designed, which is composed of three sub-controllers, i.e., cognitive controller A, PID controller and cognitive controller B. Cognitive controller A is designed with Q-learning, and cognitive controller B is designed by blending adaptive RBF neural networks with Q-learning. For an extensive analysis, the presented control methodology is compared to Q-learning-PID. The simulations show that the proposed cognitive control scheme has better robustness to packet losses and time delay than Q-learning-PID.