Numerical Methods for Model Predictive Control

This thesis mainly deals with the extended linear quadratic control problem, that is a special case of equality constrained quadratic program. A number of other problems in optimal control and estimation of linear systems can be reduced to this form. Furthermore, it arises as sub-problem in sequential quadratic programs methods and interior-point methods for the solution of optimal control and estimation in case of non-linear systems and in presence of inequality constraints. This thesis can be divided into two parts. In the first part, a number of methods for the solution of the extended linear quadratic control problem are presented and analyzed. These methods have been implemented in efficient C code and compared each other. In the second part, this problem is expanded taking into account also inequality constraints. Two interior-point methods are presented and analyzed. Both methods have been implemented in C code and compared each other