Parameter estimation using a novel nonlinear constrained sequential state estimator

A sequential state estimation routine was developed to allow for the incorporation of constraints into their estimates. In this effort, the application of this powerful state estimator to the problem where parameters of the system model are incorporated into the state vector is examined. The research has looked at the issues that arise in both the open-loop implementation, such as occurs in the target tracking application, and the closed-loop implementation that occurs in the feedback control problem. This effort is aimed toward system identification of parameters that have hard limits on their values. This type of parameter estimation can provide the foundation for a training paradigm for elliptical basis functions in neural networks.