Restricted structure predictive optimal control

The design of low-order predictive optimal controllers, that involve a multi-step cost index and future setpoint knowledge, is considered. The usual predictive controller is of high order and the aim is to develop simpler structures, suitable for applications where PID controllers might be employed. The system is assumed to be represented by a discrete-time state-space model, which is very general, and the quadratic cost-function may include dynamic cost weighting terms. Using this approach, it is straightforward to generate a much lower-order predictive controller and thereby simplify implementation. Even with a continuing improvement in computational power there are many good reasons why low-order simple controllers have advantages in real applications.