MPC RELEVANT IDENTIFICATION USING GENERALIZED ORTHONORMAL BASIS FILTERS*

In MPC relevant identification, it is necessary to identify models that are suited for multi-step ahead predictions. This can be achieved by minimizing the multi-step ahead prediction error in the identification stage. This work aims at the development of a methodology for identification of MPC relevant models based on Generalized orthonormal basis filters (GOBF). Specifically, ARX models parameterized using GOBF are identified. The efficacy of the proposed modeling technique is demonstrated by carrying out simulation studies on the benchmark Shell control problem. The relative quality of the obtained models is evaluated through closed-loop performance with MPC.