Partitioned State Algorithms for Recursive System Identification
Anthony W. Warren · 1983
In many on-line state estimation applications, identification of unknown system and measurement parameters is desired. In this paper a recursive algorithm for state estimation and parameter identification is presented which is particularly appealing due to its computational simplicity and its structure as an extension of ordinary Kalman filtering. It is assumed that the unknown states and parameters are linearizable about known reference values at each stage. The algorithm is derived as a recursive solution to a maximum likelihood estimation problem. The theory is illustrated by application to an adaptive filtering problem which arises in aircraft tracking systems.