- Report
- Heating and Cooling
Improved superheater control with feedforward from physics-based process models
| 2015:220 | MARTIN RÅBERG, HÅKAN RUNVIK, MAJED SAMMAK, OLA SLÄTTEKE, STÉPHANE VELUT
Feedforward is one of the most fundamental structures in control engineering. Anyone
who has taken a basic course in control engineering at university level should have
encountered the concept of feedforward. Feedforward is used mainly to deal with
process disturbances and traditionally it is in the simplest case a pure gain, or
alternatively dynamic linear models or static nonlinear models.
The novelty of this project is to use dynamic nonlinear physics-based multivariable models used for feedforward. The challenge of nonlinear physics-based models is the mathematical inversion required in order to use them as feedforward. The automatic methods in the modeling language Modelica to calculate and simulate inverted models has recently opened for new applications thanks to the non-causal modeling language and symbolic methods of equation manipulation. This is used in this report to produce an inverse process model in the form of a Modelica-model. This inverted process model then forms the basis for the robust and approximate inverse implemented in an industrial control system (in this case ABB Freelance).