Single neuron PID controller based on adaptive Smith predictor for networked control systems

Abstract

For networked control systems (NCS), the networkinduced time delay with uncertain nature may degrade the control performance and even destabilize the system. Therefore, it is necessary to study the strategies to compensate time delay. In order to effectively restrain the impact of network time delay, a single neuron PID controller combined with adaptive Smith predictor is proposed to achieve better compensation effect in the paper. The traditional Smith predictor is sensitive to parameter uncertainties, leading to compensation failure when the time delay is stochastic. Single neuron PID controller is adopted to improve the adaptability and the robustness of system with the ability of self-learning and the simple structure. New adaptive Smith predictor can ensure the model error converge to zero. Besides, it does not include the delay model, thus the network time delay does not need to be measured or estimated. Simulation results are given to illustrate the effectiveness and the robustness of the proposed method.

Synchronous control strategy for three-servo networked control system

“…Smith predictor by O. J. M. Smith in 1950s is able to remove the time delay from the inside of closed loop to the outside, but its use is restricted due to large dependence on precise identification model. To solve this problem, methods that Smith predictor combined with intelligent control theories, such as neuron network and fuzzy logic, are proposed in [8][9][10].…”