Simultaneous state and parameter estimation and location of sensors for distributed systems

Abstract

This paper deals with the non-linear problem of simultaneous parameter and state estimation for distributed systems, including optimal location of sensors. The class of models we are interested in is characterized by linear unbounded operators which are densely defined and dissipative. Our approach applies linear filtering techniques to a sequence of linearizations at suitable trajectories. The optimal location of sensors is carried out by minimizing a measure of the simultaneous state and parameter estimation error. This is posed as the minimization of an upper bound for the filter covariance.