Continuous Generalized predictive of an Experiment Four-Tank level Process
Loading...
Date
2016
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Obafemi Awolowo University
Abstract
The four-tank level process, a typical multi input multi output (MIMO) system exhibit complex dynamics, nonlinearities and strong interaction as seen in most industrial process plants. The complex and dynamic nature of MIMO systems, necessitate the quest for an advanced control technique that could handle strong interactions. In this study, an experimental four-tank system was designed and interface for the implementation of continuous generalized predictive control (CGPC) and other classical controllers for both minimum phase (MP) and non-minimum phase (NMP).
The designed and constructed four-tank system was modeled using mass balance and Bernoulli’s law. All components that require high precision were calibrated. The non-linear model of the system was linearized to determine the operating points. State space as well as the transfer function for the two operating phases were obtained. The Relative gain array (RGA) was used to determine a suitable decentralized input-output pairing control structure. Using input-output pairing suggested by RGA, an adopted CGPC algorithm written in MATLAB was employed to design appropriate CGPC loop controllers. Classical PID controllers based on internal model control (IMC) and auto tuning PID were parameterized and implemented on the system. These controllers were used for both the simulation and the real time experiments using MATLAB/Simulink. The controllers’ performance evaluation of the system for both MP and NMP was based on integral absolute error (IAE) and integral square error (ISE) performance indexes and other time-domain performance metrics.
The results of the experiments showed that the physical system was adequately matched with the obtained transfer functions. The steps responses obtained for both simulations and experimentations portrait that of an interacting multivariable system. Furthermore, satisfactory performances was achieved for simulation and the real time experiments with comparative lower values of IAE and ISE particularly for the MP. A gross value of ISE 13.63, 9.81 and 54.14 and IAE values of 18.61, 18.52 and 78.14 were for estimated for auto tuned PID, CGPC and IMC respectively. Also, an overall percentage overshoot of 4.91, 3.01 and 5.25% were recorded respectively for MP. On the other hand, larger value of ISE 49.48, 6267.2, 2701, and IAE values of 79.24, 4136 and 1016.4 and percentage over shoot of 0.56, 15.50 and 1.61% were obtained respectively for auto tuned PID, CGPC and IMC respectively for NMP.
The newly designed and interface four-tank experimental system will provide a platform to demonstrate and illustrate several control concepts due to its nonlinearity and complex dynamics. The system will help students to connect between control theory and the real-world thereby adding realism to control education. This study established that a satisfactory control response of the four-tank level process can be achieved with CGPC especially when operating at the MP.
Description
xvii,137
Keywords
Predictive Control, Four Tank, Level Process, MIMO, Integral absolute error, Integral square error
Citation
Olayinka,A.O(2016).Continuous Generalized predictive of an Experiment Four-Tank level Process.Obafemi Awolowo University.