Abstract

A novel concept of partial identification for Multi-Input Multi-Output (MIMO) system under input-output data based modeling is proposed in this work. Issues such as richness in identification data, plant safety, operator’s comfort, time involved in identification experiments, identifiability, model accuracy etc. needs to be addressed while conducting an identification experiment. Both open-loop and closed-loop identification methods suffer with one or the other way while considering above issues all together. The proposed partial identification framework helps in taking advantages of both open-loop and closed-loop identification setup on a common platform. The proposed method is applicable to MIMO systems and it uses direct identification method for linear system identification using input-output data. The paper also addresses the potential benefits of a nonlinear regulator/controller and dithering in partial identification framework. The simulation results on MIMO Quadruple Tank system supports the benefits suggested due to partial identification framework.

Paper Details