Citation Link: https://doi.org/10.25819/ubsi/10342
Optimalitätsbasierte Entwicklung von Zustandsschätzung und Regelung eines Fertigwalzblocks in einer Warmwalzstraße für Stabstahl
Alternate Title
Optimization based development of state estimation and control of a finishing mill in a hot rolling mill for steel bars
Source Type
Doctoral Thesis
Author
Schäfer, Marc-Simon
Subjects
Modellprädiktive Regelung
Moving-Horizon-Schätzverfahren
Warmwalzwerk
Prozesssimulation
Längsspannungen
DDC
620 Ingenieurwissenschaften und zugeordnete Tätigkeiten
GHBS-Clases
Issue Date
2022
Abstract
The requirements for industrial applications are constantly facing new challenges and the achievement of higher efficiencies. This also applies to the rolling mill industry and especially to hot rolling mills. In these, the use of sensors is only possible to a limited extent due to the harsh environmental conditions. For process optimization, however, the highest possible degree of knowledge about the plant condition is required in order to carry out optimizations on this basis.
The present work makes a contribution to this open field of development. First, the dynamic modeling of the rolling process for a finishing rolling block for steel bars is carried out. Then, optimality-based methods for estimating state variables are described and, finally, an optimality-based control system is presented. The plant states considered here mainly concern the material flow – the material velocities and cross-sectional areas as well as the longitudinal stresses between the individual rolling stands. These quantities are not available as measured variables in the considered plant layout, but provide essential insight into the process behavior.
For the mathematical modeling of the rolling process, physically based approaches are used which represent the forming behavior in the roll gap, but also the behavior of the kinematic variables. Dynamics introduced by the drive train of the rolls, and also the coupling between the rolling stands, lead to the overall system used as a simulator. These models are configured with real plant parameters and then validated with measured values from this plant.
For estimation and control, the focus is on optimality-based methods, namely Moving Horizon Estimation (MHE) and Model Predictive Control (MPC). With additional constraints and parameters from the real plant, the nonlinear optimization problems are set up, which are then solved with numerical solvers. The industrial aspect that wear and uncertainties in parameterization lead to inaccuracies in the process description is taken into account.
In this work, a rolling simulator for N rolling stands is created, which can be parameterized individually for each stand and thus covers a large number of possible configurations of finishing rolling blocks. From this, a reduced nonlinear material flow model is developed to serve as the basis for the MHE and MPC.
The MHE provides the complete state vector of the material flow, which is not available in the plant so far. In addition, roll-specific variables (e.g., the degree of deformation) and their influence on the material flow are estimated online.
The MHE/MPC approach is successfully applied to the rolling simulator and ensures constant material flow even in the presence of external, unknown disturbances. It is shown that for known disturbances, the settling time is reduced by up to 20 %. In addition, it is shown in simulation that the use of sensors for the input variables, represents a further performance gain. In comparison with conventional controller structures, it is shown that the consideration of process limitations and measurement of external disturbance variables leads to a significant reduction of the settling time.
Finally, an example is given of how the determined solutions can be reformulated for real-time operation, and an overview is given of the different runtime durations of the optimality-based methods.
The present work makes a contribution to this open field of development. First, the dynamic modeling of the rolling process for a finishing rolling block for steel bars is carried out. Then, optimality-based methods for estimating state variables are described and, finally, an optimality-based control system is presented. The plant states considered here mainly concern the material flow – the material velocities and cross-sectional areas as well as the longitudinal stresses between the individual rolling stands. These quantities are not available as measured variables in the considered plant layout, but provide essential insight into the process behavior.
For the mathematical modeling of the rolling process, physically based approaches are used which represent the forming behavior in the roll gap, but also the behavior of the kinematic variables. Dynamics introduced by the drive train of the rolls, and also the coupling between the rolling stands, lead to the overall system used as a simulator. These models are configured with real plant parameters and then validated with measured values from this plant.
For estimation and control, the focus is on optimality-based methods, namely Moving Horizon Estimation (MHE) and Model Predictive Control (MPC). With additional constraints and parameters from the real plant, the nonlinear optimization problems are set up, which are then solved with numerical solvers. The industrial aspect that wear and uncertainties in parameterization lead to inaccuracies in the process description is taken into account.
In this work, a rolling simulator for N rolling stands is created, which can be parameterized individually for each stand and thus covers a large number of possible configurations of finishing rolling blocks. From this, a reduced nonlinear material flow model is developed to serve as the basis for the MHE and MPC.
The MHE provides the complete state vector of the material flow, which is not available in the plant so far. In addition, roll-specific variables (e.g., the degree of deformation) and their influence on the material flow are estimated online.
The MHE/MPC approach is successfully applied to the rolling simulator and ensures constant material flow even in the presence of external, unknown disturbances. It is shown that for known disturbances, the settling time is reduced by up to 20 %. In addition, it is shown in simulation that the use of sensors for the input variables, represents a further performance gain. In comparison with conventional controller structures, it is shown that the consideration of process limitations and measurement of external disturbance variables leads to a significant reduction of the settling time.
Finally, an example is given of how the determined solutions can be reformulated for real-time operation, and an overview is given of the different runtime durations of the optimality-based methods.
File(s)![Thumbnail Image]()
Loading...
Name
Dissertation_Schaefer_Marc_Simon.pdf
Size
6.6 MB
Format
Adobe PDF
Checksum
(MD5):060aee586935618298d1d0d65e3328ca
Owning collection