Sensitivity Compensating Control: Data-driven model based adaptive approach

A novel Sensitivity Compensating Control (SCC) approach is proposed in a data-driven model based platform and combined with an Extended External Reset Feedback (EERF) method to handle sensitivity, input saturation, and accurate process model requirement problems associated with application of Generic Model Control (GMC). Two versions of Adaptive GMC (AGMC) are proposed using linear-in-parameters time-series models with time-varying parameters for higher relative degree systems, and are used in the formulation of SCC and EERF approaches. The steps involved in the proposed approach consist of defining a new process, control law and set point such that the determined control action drives the original process to its desired set point. The performance of the proposed control algorithms is illustrated by application to a benchmark multi-product polymerization reactor control challenge problem. The proposed approaches are applicable to chemical engineering systems exhibiting input sensitivity.

► A novel data-driven model based Sensitivity Compensating Control approach proposed.
► Two versions of adaptive GMC proposed using linear-in-parameters time-series models.
► Involves definition of new process, control law and set point to achieve smooth inputs.
► Approaches tested on a multi-product polymerization reactor control problem.
► SCAGMC-I found to exhibit better performance over other controllers.