Controle PID e preditivo: aspectos teóricos e de implantação prática

Abstract

Nowadays, we observe the exponential increasing of demand for performance optimization in all areas of modern life. This search for better performances is also occurring in all industry sectors, either factory or energy activities. This best performances reflects an increase is all levels and achieve enormously the process control systems. With more strict performance requirement, technique formerly studied only in academic circle has been migrate for industry area, opening new horizons and establishing new concepts in general. It is not different with the process control systems, even in particular at the lowest level of implementation (factory’s floor) has been noted a constantly update. This paper is inserted into this last concept, in other words, into the project of loop control process at lowest level of the plant. Thus the aspect of tuning of conventional multivariable like PID (Proportional, Integral, Derivative) controllers (MIMO) and MIMO predictive controllers based on models like GPC (Generalized Predictive Control) type are studied aiming to the practical implementation. Several system control aspects, such as identification of process through parametric and nonparametric technical with different representations and aspects of decoupling between the process loop, as it is MIMO system, are also dealt, discussed and implemented. For that purpose, two plants were developed at IFPI’s NEE (Núcleo de Engenharia Elétrica) laboratory. The first platform consists of MIMO system of ventilation and heating; the second platform is a system of coupled double tanks. As a conclusion of the study of practical experiments conducted, can be established that the best control for a process depends on the application for which it is destined and for the proper functioning is necessary the theoretical and practical knowledge of the dynamic of the process. Thus an appropriated controller depends also on the use of a good identification technique.