A new generalized model-based predictive control algorithm

A unified generalized model-based predictive control (GMBPC) technique is presented. This technique combines in an efficient way the key properties of several previous MBPC-like algorithms. The multiple-input-multiple-output (MIMO) state-space is employed, and state and control constraints are included in the system formulation. For better accuracy a second-order model is employed for each output variable, while a first-order model is always used in the available algorithms. For the unconstrained case a simple explicit control law is obtained. A particular feature of the proposed technique is that the predictive functional control principle is used to reduce the computational complexity of the resulting GMBP controller. Extensive simulation examples in industrial and managerial system models support the effectiveness of the present GMBP controller in terms of meeting the desired specifications and having reduced computational requirements.<>