CONTROL SYSTEM DESIGN FOR MULTIVARIABLE NONLINEAR SYSTEMS GUARANTEEING GOOD PERFORMANCE OVER WIDE OPERATING RANGES

It is sometimes necessary that a process plant operates over a wide range while satisfying desired performance and stipulated constraints. This paper proposes a method based on computing a mean (nominal) plant, uncertainty weight based on the deviations of the extreme plants from the mean plant and performance weights capturing desired performance. The work considers simple multi-loop and centralized controllers and their parameter determination using optimization. In order to expeditiously compute the centralized controller parameters, its initial parameters are determined using the parameterization based on the multivariable internal model controller (MIMC) of the p/q Pade approximant of the mean plant. The order of magnitude of the MIMC filter is determined by noting the magnitude of a suitably computed multiloop controller which, at least, stabilizes the closed loop system. In this work, desired metrics for the closed loop system include robust performance, minimization of the closed loop system integral of squared error (ISE) and limiting the maximum controller outputs over the full operating range. It is demonstrated that the new method facilitates the computation of simple centralized controllers yielding closed loop systems with favorable characteristics.