Direct recursive estimation of noise statistics
Abstract
A Kalman filter requires an exact knowledge of the noise statistics. The problem of identifying the noise statistics was considered. A recursive algorithm for the estimation of the covariances and means of the state and observation noises is presented. The estimated values coverge to their true values in the mean square sense. One of the features of the presented method is that it does not require the state estimation. The stationarity is not assumed for any kind of obtainable data. The identification of the covariances can be done independently from that of the means of noises. Specifically, the covariance of the observation noise can be done without being affected by the other unknown factors of the noise statistics. Explicit formulae, which the means of the state and measurement noises must statisfy, are presented.
- Publication:
-
Ph.D. Thesis
- Pub Date:
- April 1977
- Bibcode:
- 1977PhDT........29O
- Keywords:
-
- Noise Measurement;
- Recursive Functions;
- Statistical Distributions;
- Algorithms;
- Amplification;
- Covariance;
- Kalman Filters;
- State Estimation;
- Electronics and Electrical Engineering