Gradient estimation algorithms for equation error formulations

Abstract

This paper presents the theory for and illustrates the application of gradient parameter estimation algorithms which have been developed for equation error formulations of parameter identification problems. These algorithms are computationally simple; hence, they are ideal candidates for on-line applications. They are limited, however, by availability of first- and second-order statistics of noise processes.