Control or predictive models based on Model Predictive Control (MPC) does not refer specifically to the design of a controller but rather a set of ideas or features for the development of control strategies, applied in a greater or lesser degree, give rise to different types of drivers with similar structures. The MPC is one of the techniques of control that has developed in the academic and industrial in recent decades due to its simplicity and efficiency. 

However, it is not easy to link the adjustment of controller parameters and performances of the closed loop. In this regard, it is important to design algorithms for predictive control to ensure the stability of the nominal closed loop, with small computing time and with a clear meaning of its parameters on the performance of the system or on the control effort. 

The major contribution of this thesis is related to the definition of a new type of predictive controller, the PC-GPC, a modified version of standard GPC. This controller has been replaced the weight of the control action for a new parameter called the number of principal components (NPC). The relationship between the new parameter (NPC) and some numerical indicators, such as the standard vector control activities or the number of dynamic condition of the matrix G, make your choice based on subjective criteria unless the balance of the shares control. Furthermore, it has tested this type of controller in both SISO and MIMO processes and their characteristics of robustness and stability. On the other hand, has drawn a method for calculating a PC-GPC controller to ensure stability of the nominal closed loop, when the model is known exactly.