Computationally efficient globally linearizing control of a CSTR and the Tennessee Eastman problem using quadratic perturbation models

This work aims at the development of computationally efficient schemes under the discrete time multivariable globally linearized control (GLC) framework. Unconstrained and constrained GLC schemes are developed using a discrete quadratic perturbation model. The structure of QPM facilitates analytical treatment of the unconstrained controller synthesis problem for square multi-input multi-output (MIMO) nonlinear processes and makes it possible to develop closed form nonlinear control law. For non-square MIMO systems and for handling input constraints, an optimization based discrete GLC formulation is developed. The effectiveness of the proposed GLC formulations is demonstrated by conducting simulation studies on a CSTR system and the benchmark Tennessee Eastman (TE) control problem.