Design of the Adaptive Smith Predictor for the Time-Varying Network Control System

Abstract

Ibis paper integrates Ethernet and control area network (CAN) with the TCP/IP communication gateway. The developed system has been applied to a remote control system for an AC 400W servo motor to verify the proposed adaptive Smith Predictor control method with the time-delay estimation algorithm. The time delay effect, defined as the round-trip time (1117), between the application layer of the server and the application layer of the client, has been measured and analyzed. Moreover, results indicate that NCS performance is degraded, and may even become unstable mainly due to the serious network time delay in a stochastic nature mainly occurred in the Ethernet. The Smith predictor is adopted to eliminate the constant time delay. However, the delay in a commercial network is time varying and it depends on the number of end users. An adaptive Smith predictor control scheme is developed in this paper by applying the on-line estimation to further improve control performance of the proposed NCS with the integrated Ethernet and CAN buses. Experimental results indicate that the time delay is drifting in both bounded and random natures for the integrated Ethernet+CAN network and the proposed control approach significantly improves motion accuracy as in the NCS for an AC motor-based motion system.