Çok girişli çok çıkışlı kısıtlamalı sistemlerin öngörülü kontrol uygulamaları

Abstract

PREDICTIVE CONTROL APPLICATIONS OF MULTI INPUT MULTI OUTPUT CONSTRAINED SYSTEMS Recently, multivariable control techniques have received increased industrial interest especially in chemical industry. But various problems are encountered in multivariable control applications since most industrial process have multi input and multi output with interactions in between. Such systems are called Multi Input Multi Output (MIMO) systems. Also physical constraints exist in MIMO systems. These constraints can be on the input due to actuator limitation as well as on the outputs and also can be due to safety, product quality and efficiency requirements. In any control application it is important to take these constraints into consideration. In the conventional controllers each variable is controlled separately such as pressure, temperature, liquid level, flow etc. This is not ideal solution for the process automation. It is desired to design a single controller to control all the variable simultaneously. To handle the above mentioned problems predictive control algorithm which is based on the prediction of the system output has become a major research area. In this study, we first analyses of the MIMO systems and information on some multivariable algorithms were presented. A classical Quadruple Tank application was performed using classical multivariable trajectory tracking PI controller to illustrate the concepts in multivariable control. A pressure-liquid level system which has two input and two outputs (TITO) has been designed and build on which predictive control has been applied. A mathematical model for the process system based on experimental results was derived and tested. Transfer function parameters calculated in real-time using the input-output relation. The open loop response of the obtained model was compared to the real-time open loop response of the system. Information of model predictive control algorithms were given, how these algorithms were expanded for MIMO systems and the how constraints were incorporated was explained. Due to the structure of predictive control, constraints have been easily defined and model predictive control algorithm has been easily applied to our pressure-level system. The designed results of the controller were compared with results of decoupling PI which is a classical multivariable control technique, experimental and simulation results were presented. The experimental results show that model predictive controller (MPC) controls the pressure-level tank system in desired limits. When compared with PI controller also satisfactorily controls the system as MPC under variable references. But when the constraints were considered the MPC controller works better than PI controller.