抄録
This paper is divided into two parts. The first is concerned with the performance loss of the Wiener-Kalman filter (or the Bayesian estimation process) due to incorrect information associated with the values of covariances of system noise and of observation noise. By evaluating the actual, calculated and optimal covariance matrices, the difference equations are derived to describe evolution of three error matrices. It is clearly shown for the first order system that the Wiener-Kalman filter (or the Bayesian estimation process) is rather insensitive to variations in noise levels.
The remainder of this paper is devoted to the investigation of the mutual relations between the estimation problem and the regulator problem from the viewpoint of performance loss. It is well known that the estimation problem is the dual of the noise free regulator problem. We demonstrate here that there still exists the duality principle between the two problems in the sense of performance loss.
