PMF method in the presence of noise

Abstract

The problem of parameter estimation in a wide class of problems of lumped continuous systems is tackled through the well established PMF technique. Dynamic models including a class of nonlinear and time-varying cases can be handled without any difficulty by the PMF method if the models are linear in their parameters. In a noisy environment, the system of PFC may be augmented with well defined Kalman filters for optimal estimation of the PMF's of the process data. An advantageous feature of the present method is that the filtering problem is independent of the type of model under identification. It depends only on the well defined linear PFC system. In an unknown environment the technique of adaptive Kalman filtering [G1, G19], seems to be applicable. With this the PMF method of system identification appears to achieve the desirable generality, thereby becoming a very dependable approach to continuous sytem identification.