A Simple Means of Updating the SRIF Filter When the State Equations are in Triangular Form.

摘要

The square-root information filter (SRIF), which is a numerical method of implementing the Kalman filter, was investigated. Under an important special case of no process noise and an upper triangular form of the state transition matrix, the prediction process of the filter is of simple form. The transformed smoothed and predicted states are identical, and the smoothed and predicted covariance factors are related by a simple matrix transform. (Author)