State estimation with intermittently interrupted observation
The optimal estimation of the system state is studied where the observables are intermittently interrupted. The primary aim of this research is to determine the more versatile filtering algorithm which can be applied to various systems with quite different system requirements.
To deal with the two assumptions; (1) the correct signal statistics are not available, and (2) the statistical property of the interruption function (γ(k)) is not known, a new concept which incorporates both decision theory and Kalman Filter is developed.
The interruption is treated as a switching observation error variance in the observation equation. Then the structure of the underlying stochastic process may change between two possible forms. The best estimation algorithm is selected according to this environmental condition. The adaption is dictated by a threshhold which is established based on the statistical decision theory. Two decision making approaches are proposed. The formulas for calculating the threshhold level and the estimation error variance are developed. And, possible degradation region of the filter performance with the interruption is investigated.
The theoretical developments are confirmed by computer simulation of several sample systems. The theory can be applied to the optimal estimation of the system with the interrupted observation.
Thesis78b.T353.pdf
3.17 MB
Adobe PDF
e987aead6c740938dfa3a3a849c9e183