ABSTRACT
This chapter focuses on how the Model Predictive Control (MPC) algorithm itself can be modified to reduce computational load. Predictive control strategies allow for the systematic handling of constraint, performance, and stability. However, the associated algorithms can be computational burdensome and/or difficult to unravel. The main price of computational simplicity is a reduction in the size of the feasible region. This is because there are less d.o.f. to give freedom in the predictions. There will also be some suboptimality due to the implied restriction in the parameterisation of the future control trajectory. NESTED shows a distinct change in control and therefore suboptimality. Efficient MPC and ONEDOF are often able to come arbitrarily closed to the global optimal with a negligible computational burden. The most common method for solving a quadratic programming (QP) is the active set method. The QP algorithm is available within MATLAB and hence readily accessible to researchers as a tool.
