The minimization of measurement error in a general perturbation-correlation process identification system

A general identification system is studied for an important class of realistic, time-varying processes. This class consists of those in which the process is nominally known, and the statistical characteristics of its varying parameters and of the environment are also known. The expression for identification error in terms of the spectral properties of the parameter variations and of the output transducer noise is developed. Optimization procedures are given to minimize the perturbation-correlation system's mean-square identification error.