Subspace predictive control. / Controle preditivo com enfoque em subespaços.

AUTOR(ES)

Erika Maria Francischinelli Fernandez

DATA DE PUBLICAÇÃO

2009

RESUMO

Model Predictive Control (MPC) technology is widely used in chemical process industries. Subspace identification (SID) on the other hand has proven to be an efficient alternative for classical system identification methods. Based on the results from MPC and SID, it was developed in the late 90s a new control approach, called Subspace Predictive Control (SPC). This approach is also known as data-driven predictive control. In this new method, one single operation replaces the three steps in a MPC controller design: system identification, the state observer design and the predictor matrices construction. The aim of this work is to review studies in the field of SPC, to apply this technology to typical systems of chemical industry and to propose new algorithms. It is developed three internal excitation algorithms for the SPC method, which allow the system to be persistently excited while a minimal control of the process is still guaranteed. These algorithms enable the application of closedloop identification, where the SPC controller model is re-identified using the previously excited data. The SPC controller and the SPC controller with internal excitation are tested through simulation for two different processes. The first one is a debutanizer column of a distillation unit for which two linear models corresponding to two different operating points are available. The second one is a non-linear system consisting of a styrene polymerization reactor. A phenomenological model is provided for this system. Tests results indicate that SPC is more susceptible to measurement noises. However, it is noticed that SPC controller corrects perturbations on set-points faster than MPC. Simulations for the SPC with internal excitation show that the proposed algorithms sufficiently excite the system, in the sense that more precise models are obtained from the re-identification with excited data.

ASSUNTO(S)

identificação em malha fechada subspace identification controle preditivo orientado a dados controle preditivo closed-loop identification subspace predictive control data-driven predictive control process control predictive control controle preditivo com enfoque em subespaços identificação em subespaços controle de processos

Documentos Relacionados

• Integração da otimização em tempo real com controle preditivo.
• Modelagem e controle preditivo utilizando multimodelos
• Model predictive control for Francis hydro turbine generators
• Controle preditivo de horizonte infinito para sistemas integradores e com tempo morto.
• Neural predictive control applied to aerospace vehicles