DMC filtrado aplicado ao controle de sistemas sujeitos a referências variantes no tempo

Abstract

The Model Predictive Control (MPC) strategies are well-known for their ability to naturally handle future reference information. However, in the presence of time-varying references, mitigating the reference tracking error may require potentially aggressive MPC tuning in the context of control effort. The Dynamic Matrix Control (DMC) is widely used due to the convolutional model, which is directly defined from the step response. The inherent robustness of the original DMC is defined by the same design parameters that define the reference tracking performance. In this work, the filtered DMC is used to balance the trade-off between tracking error performance and robustness/noise attenuation properties. A simulation example based on the temperature control of an NTC sensor and an experimental case study are used to illustrate the benefits of the filtered DMC in the context of tracking error in the presence of time-varying references.