Internal model pre-processing tool

Abstract

In industry the model predictive control (MPC) solutions are for their high computational requirements usually implemented to the PC-based devices. The main goal of our study is to implement an MPC algorithm to the programmable logic controller (PLC). This paper describes an initialization part of our solution, which represents an internal model pre-processing algorithm implemented to the function blocks created in PLC programming standard IEC 61131-3. The pre-processing algorithm changes model representation form of multivariable system (MIMO) from continuous transfer function matrix to generalized discrete state-space model. At the end of the paper there is shown created comfortable interface used for inserting of transfer functions, process model parameterization and future MPC tuning. This interface communicating with pre-processing algorithm running on PLC allows to specify MIMO internal model by continuous transfer function matrix and the discrete state-space internal model used during control is created automatically. The operator does not have to use MATLAB or other expansive mathematical software for creation of discrete state-space representation and the most widespread mathematical description in industry known like continuous transfer function can be used. The constants of transfer functions are usually obtained by operator from current visualizations on the plant.