Unlike other sensors, process tomography can really account for the fact that many industrial processes are distributed parameters systems. However, appropriate automatic control methods are mostly still to be developed. Reports on control applications of tomography are rare and the proposed approaches were almost never physically demonstrated. The objective of the project is to develop automatic control method(-s) that will be appropriate for tomography data. When developing this method, it will be necessary to consider the fact that the image reconstruction is not only computationally demanding but also ill-posed. Consequently, approaches that use raw data and that are based on suitable process model parametrization or state estimation are more likely to succeed than the approaches that strictly separate the image reconstruction and control. The method must take into account the significant uncertainty in tomography data and finally it shall be implemented in real time and tested with the continuous casting process using the Mini-LIMMCAST facility available at HZDR.
Research Activities / Objectives:
- Analyze the application areas of control based on process tomography
- Propose and develop an appropriate automatic control method(s) that is suitable to be used in connection with process tomography in the role of sensor of controlled variable(s)
- Develop concepts that will allow to avoid the image reconstruction phase and to use the raw data instead
- Develop approaches for the description and handling of uncertainty in tomography data and use them to make the controller sufficiently robust
- Apply the developed controller concept to continuous casting process and demonstrate its functionality and performance using the Mini-LIMMCAST facility at HZDR