Artificial Neural Network-based Smith Predictor for Compensating Random Time Delays Acting in Networked Control Systems

Abstract

Networked Control Systems (NCSs) denote a kind of controlled system in which, process, sensors, and controllers share information and signals through a limited bandwidth network. The presence of a network in an NCS’s control loop causes delays, which may often lead to many imperfections, such as packet losses, multi-packet transmissions, packet disordering, or even system instability. This paper proposes an artificial neural network (ANN)-based Smith predictor to compensate for delays in NCSs. Two ANNs are used: one for estimating a model for the free-delay plant and the other for approximating the time delays affecting the system. The validity and effectiveness of the proposed approach are shown through a simulation example in MATLAB®/Simulink and TrueTime.