Model Predictive Control Using Orthonormal Basis Filter

Proportional Integral Derivative (PID) controller is the most common controller that
acts as standard tool in a process control industry. However, when interacting with
Multiple Input and Multiple Output (MIMO) process, the interaction is difficult to be
controlled by PID controller. Therefore, this project will focus on Model Predictive
Control (MPC) that is one of optimization strategy that can control MIMO interaction
by predicting the effect of potential control action. In this project, a mathematical
model of Orthonormal Basis Filter (OBF) will be developed on the distillation column
based on Wood-Berry model with a feedback control (a closed loop system). A
simulation of MPC is done by using MATLAB coding while PID is simulated using
SIMULINK. Based on the simulation, the performance of MPC and PID controller are
evaluated by using the Integral Error Criteria: Integral Absolute Error (IAE), Integral
of the Squared Error (ISE) and Integral of the time-weighted absolute error (ITAE)
and also with total input variation. Lower integral error criteria and total input variation
value indicate a better model accuracy and efficiency of controller for MIMO system.