Non-liner model predictive control of a multivariable process

Abstract

<P>MOST MODEL PREDICTIVE CONTROL (MPC) IMPLEMENTATIONS IN PROCESS INDUSTRY ARE BASED ON LINEAR MODELS. THEY ARE NOT SATISFACTORY FOR OPERATION AWAY FROM NOMINAL STEADY STATE OR WHEN PROCESS IS NONLINEAR. THIS LED TO THE DEVELOPMENT OF NONLINEAR MPC (NMPC). VERY FEW OF THE MANY STUDIES THAT HAVE BEEN REPORTED ON NMPC ARE EXPERIMENTAL AND NO STUDY OF NMPC FOR PH AND LEVEL CONTROL OF A NEUTRALIZATION PROCESS HAS BEEN REPORTED. HENCE, IN THIS WORK, NMPC AND A COMPUTATIONALLY EFFICIENT NMPC HAVE BEEN DEVELOPED AND APPLIED TO PH AND LEVEL CONTROL VIA SIMULATION AND EXPERIMENTATION. ANOTHER TECHNIQUE, NONLINEAR DYNAMIC MATRIX CONTROL (NONLINEAR DMC), AND A MULTI-LOOP PI CONTROLLER ARE ALSO DESIGNED FOR COMPARISON WITH NMPC CONTROLLERS. MATLAB AND LABVIEW ARE SUCCESSFULLY EMPLOYED FOR IMPLEMENTING THE FOUR CONTROLLERS AND THEIR PERFORMANCE IS TESTED FOR BOTH SERVO AND REGULATORY CONTROL. RESULTS ARE PRESENTED AND THE RELATIVE MERITS OF THE FOUR CONTROLLERS ARE DISCUSSED.</P>