Use of Information on the Difference between the Outputs of a Plant and Its Model to Improve the Control System Performance.

The paper presents a control algorithm for the linear-quadratic problem, making it possible to improve the system performance in spite of model inaccuracy. The algorithm consists in the continuous measurement of difference between the model and plant outputs and its extrapolation. An extrapolated signal forms the basis for determining epsilon - optimal control (as a result open-loop or closed loop system can be obtained). A way to estimate the extrapolation interval under constraints imposed upon the rate of error variations is proposed.