Simultaneous identification of time-varying parameters and estimation of system states using iterative learning observers

This paper presents the design of an Iterative Learning Observer (ILO) for the purpose of estimating system states while simultaneously identifying timevarying parameters. The proposed ILO uses a novel updating mechanism to identify time-varying parameters instead of using integrators which are commonly used in classical adaptive observers to identify constant parameters while estimating system states. The main idea behind the design of the ILO is the use of learning, i.e. previous information is combined into the ILO for identifying online timevarying parameters. Stability of estimation error dynamics and convergence of parameter estimation error are established and proven. An illustrative example exhibits the effectiveness of the ILO. Copyright c ©2005 IFAC