ABSTRACT
This chapter gives a brief summary of a standard predictive control algorithm. It provides an overview of the main components in a predictive control law. The chapter explores how to ensure unbiased estimates of the steady state values by using an appropriate disturbance model in the observer. In the event of infeasibility of constraints, the soft constraints must be relaxed to ensure the Model Predictive Control (MPC) algorithm has a well-posed optimisation. This procedure is nongeneric and may often be dealt with at a higher level rather than in the MPC algorithm. Classical MPC algorithms use the future control increments as the degrees of freedom in the optimisation. MPC algorithms are better able to handle complexity because they make more use of model information, the price of course being a higher load and a less transparent algorithm.
