MODEL PREDICTIVE CONTROL OF A TUBULAR AMMONIA REACTOR-HEAT EXCHANGER SYSTEM WITH SQUARE AND NON-SQUARE CONTROL STRUCTURES USING SINGULAR VALUE DECOMPOSITION (SVD)

Part I: Development of the MPC Algorithm and SVD Computation

Abstract

This paper presents a multivariable model predictive control design strategy which incorporates manipulated variable blocking, and its computation using the singular value decomposition (SVD) approach. The attraction of the SVD approach is that it is a more reliable and transparent method of computing the solution of unconstrained least-squares optimization problems which often become ill-conditioned for typical tuning and design parameters employed in model predictive control. Guidelines for the application of the SVD approach are presented and discussed.