Performance-Index Functions in Networked Control Systems With Disturbance and Noise

Abstract

This paper presents performance-index functions (PIFs) in networked control systems (NCSs) with disturbance and noise using nonlinear approximations. Based on experimental data, exponential and polynomial approximations are formulated to describe the system performance versus frequency and disturbance. Approximation methods can be used to estimate the amount of disturbance at a given frequency. Once the frequency and the magnitude of disturbance are determined, the optimal sampling frequency can be calculated from PIFs. The exponential and polynomial approximation techniques can be used for an NCS to run within an allowable performance level using PIFs with minimizing the system bandwidth utilization (BU). In this paper, a DC motor speed control system through network is used as an example. From experimental data, the coefficients for exponential and polynomial approximation equations are calculated using a trust-region algorithm and a linear least squares algorithm, respectively. Although the exponential method traces the experimental data better than the polynomial one, it will also take up more resources in real time and may degrade the NCS performance if its calculation time takes more than the allowable time for a given sampling period. Thus a balance between cost and performance should be maintained.