An analysis of Variance Approach to tuning of Generalized Predictive Controllers for Second Order plus Dead Time models

摘要

This paper presents a new tuning strategy for Generalized Predictive Controllers (GPC) based on Analysis of Variance (ANOVA). This strategy is derived for Second Order plus Dead Time (SOPDT) models of an industrial plant. Moreover, SOPDT modeling allows oscillating modes to be included in the model dynamics. The tuning strategy employs a simple expression for the tuning parameter as a function of plant parameters which is absent in earlier tuning attempts. This novel expression is extracted using ANOVA method combined with nonlinear regression. Also, a better performance index value and more convenient implementation are obtained in comparison with the conventional GPC tuning methods. Therefore, the tuning strategy for SOPDT models is both more comprehensive and more effective than traditional First Order plus Dead Time (FOPDT) model tunings. The proposed strategy is verified by two comparative simulation studies.