DESIGN OF IDENTIFICATION EXPERIMENTS FOR ROBUST-CONTROL - A GEOMETRICAPPROACH FOR BIVARIATE PROCESSES

A new approach to the design of experiments is proposed to identify li near multiple-input-multiple-output (MIMO) models that will provide ro bust control. The experimental designs for identification are based on minimizing uncertainties in the structure of the multivariate model r ather than simply the magnitude of identification error. The experimen tal designs are derived for steady-state robust stability of 2 x 2 sys tems using a geometric approach. This approach leads to a simple and u nified design approach based on the singular value decomposition (SVD) of the process gain matrix. These robust or control-relevant identifi cation designs for MIMO systems differ considerably from traditional d esigns developed for single-output systems. Typically in these new mul tivariate designs, the inputs are correlated, they are not binary sequ ences, and the magnitudes of the perturbations in low-gain directions are much larger than those in high-gain directions. The results are ex tended to identification under closed-loop conditions. Dual compositio n control of distillation processes is used to illustrate the physical interpretations and the effectiveness of the SVD-based design.