Performance assessment based on minimum entropy of feedback control loops

Abstract

Assessing the quality of existing industrial control loops is becoming an important routine auditing task for the control engineer. While variance is usually as an index implied in most researches and commercial activities of performance assessment methods for feedback control loops. The general assessment methods are quite effective in assessment of linear Gaussian system, however when the disturbance is subjected to non-Gaussian distribution such as bimodal distribution, performance assessment based on variance may be unreasonable. In this paper, to assess non-Gaussian control loop system, entropy of feedback-invariant in linear non-Gaussian system is mentioned firstly. Then, according to a new benchmark based on information theory and minimum entropy criterion, and the autoregressive method of estimating entropy index is proposed. Finally, examples are given to demonstrate the effectiveness and accuracy.