A control-relevant multivariable system identification methodology based on orthogonal multifrequency input perturbations

In this paper, we propose a MIMO system identification method utilizing an orthogonal multifrequency input perturbation which we call the "zippered" Schroeder-phased signal. Guidelines for the design of a plant-friendly "zippered" Schroeder-phased input are presented based on a priori knowledge of open-loop time constants and closed-loop system response requirements of the system to be identified. A frequency response estimate is then obtained by applying discrete Fourier transforms (DFTs) to the data. Control-relevant parameter estimation via a frequency-weighted curvefitting problem formulation tailored for orthogonal signal perturbations is presented. Application of the methodology is shown for the case of Model Predictive Control of a high purity binary distillation column, a nonlinear, highly interactive system.