Lyapunov-based nonlinear observer, design for stochastic systems

Abstract

An observer design methodology which is applicable to more general nonlinear stochastic system models is given. The method relies not on the optimization theory but on Lyapunov-type stochastic stability results which can guarantee a mean square exponential rate of convergence for the estimation error. It is proved that discrete- and continuous-time state estimation is possible using the method. An example is given to illustrate the performance of this observer relative to some of the most commonly used filters in this field.<>