Rapid Estimation by Detecting Probabilistically Unkown Impulse Inputs,
Abstract
In the application of Kalman Filtering for state or parameter estimation, besides the exact mathematical model of the system, the knowledge of the statistics of various random processes, such as the input disturbance and measurement noise, is essential for the proper convergence of the filtering. Large error in any one of them can cause the filter to diverge. In practical applications, an item often missing from the list of known's is the statistics of the input. If the input is a random process, then the statistics may be estimated by methods suggested by various authors. The problem of estimating the parameters (or state) of a system becomes more acute if the estimation is performed on line, because then one cannot recycle the data. In a tactical situation, the input has to be estimated extremely fast to get a quick convergence of the filter. The paper considers the problem of estimating the parameters (or state) of a system that is known to be subject to an impulse type input as may be the case for a satellite that maneuvers in space. (Author)
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1972
- Accession Number
- AD0754013
Entities
People
- C. N. Shen
- P. Sanyal
Organizations
- Rensselaer Polytechnic Institute