Recursive prediction error methods for adaptive estimation

Abstract

Convenient prediction error methods for identification and adaptive state estimation are proposed and the convergence of the recursive prediction error methods to achieve off-line prediction error minimization solutions studied. To set the recursive prediction error algorithms in another perspective, specializations are derived from significant simplifications to a class of extended Kalman filters. The latter are designed for linear state space models with the unknown parameters augmenting the state vector and in such a way as to yield good convergence properties. Also specializations to approximate maximum likelihood recursions, and connections to the extended least squares algorithms are noted.