An online parameter estimator for quick convergence and for time-varying linear systems

A recursive algorithm called 3-OM is presented to estimate parameters and process noise variances for systems in state space form. 3-OM is globally convergent to minima of the asymptotic negative log likelihood function. 3-OM approximates the quick convergence attained by the optimal nonlinear filter used as a parameter estimator. In the present form, 3-OM must be coded by the user, because a third-order moment update equation must be algebraically simplified beforehand by a symbolic processor to obtain a tractable computational load while running 3-OM.