Abstract
The square-root method is applied to the LQG problem that derives the optimal estimate and control using a quadratic performance index in large-scale linear systems.
First a square-root solution of the regulator problem similar to that for the Kalman filter is derived. Then it is combined with the square-root solution of the estimation problem and a unified approach to the LQG problem is presented. By means of this square-root scheme, significant enhancement of accuracy in numerical computation is expected and the stability of the solution of Riccati type equations in estimation and control problems will be guaranteed.
