Sensitivity analysis for parameter identification using optimal control indices

A systematic procedure has been developed for selecting, for identification, those parameters that have the most significant effects on the system performance. This is accomplished via a novel sensitivity analysis using a linear quadratic (LQ) performance index to select the most sensitive parameters and to thus pare down the number of parameters to be identified. This is necessary because identifying too many parameters may lead to poor estimates. Nonlinear optimization methods are subsequently used for identifying the sensitive parameters. For validation, a linear quadratic regulator (LQR) is designed based on this identified model and its responses are compared with those corresponding to an LQR designed for the true system