Control Relevant Dynamic Modeling and Model-based Conrol using GOBF-ANN Models

A combination of transfer functions (generalized orthonormal basis filter networks or GOBF) and artificial neural networks (ANN) has been investigated recently in our research group, and a Matlab toolbox has been developed for constructing block-oriented   nonlinear dynamic models using experimental data generated by exciting a plant. This project involves using the toolbox for the development of control-relevant GOBF-ANN based dynamic models of the following experimental setups available in the Automation Lab: (i) Heater-Mixer setup and (ii) Packed bed distillation column. These models will be further used to implement model-based control schemes such as Nonlinear Model Predictive (NMPC) or Nonlinear Internal Model Control (NIMC) on the experimental setups.