New model-based double-loop iterative strategy for integrated system optimization and parameter estimation of large-scale industrial processes

Abstract

A novel hierarchical integrated system optimization and parameter estimation technique is described which determines the optimum steady state operation of an interconnected industrial process in spite of deficiencies in the model. The technique is iterative and involves successive solutions of system optimization and model parameter estimation problems, utilizing information feedback from the real process. Particular emphasis is given to a strategy where the coordination task is divided into two nested iterative loops, with the inner loop optimization containing a self-adaptive model. Derivatives of real process measurements required by the outer loop iteration are also used to update the model. New modifiers are introduced to cater for the output-dependent constraints. This model-based double-loop iterative strategy retains an important practical advantage in that it reduces the required number of set point changes to the real process. Optimality properties and convergence conditions are investigated. A simulation study is also presented.