Details
| Original language | English |
|---|---|
| Pages (from-to) | 351-356 |
| Number of pages | 6 |
| Journal | IEEE Control Systems Letters |
| Volume | 2 |
| Issue number | 3 |
| Publication status | Published - Jul 2018 |
Abstract
Nonlinear thermal simulations of distributed parameter systems with complex geometry can be performed using finite element analysis (FEA). In order to achieve accurate results, fine spatial and time discretization is required, which often leads to large computation times. However, many methods from system theory, such as parameter identification, real-time model-based control, and model-in-the-loop simulation, heavily rely on either multiple iterations or computation time limits. Hence, a direct model deviation from FEA is unfeasible for these approaches. Model order reduction (MOR) techniques have been proposed to improve computational performance. However, most of them are only applicable to linear systems, but linearization of nonlinear boundary conditions over a wide temperature range does not always fulfill accuracy requirements. Therefore, we propose a simplified nonlinear system description by decoupling nonlinear affected states, performing MOR of the remaining linear term and apply calculated projection to the nonlinear affected part. During simulation, the reduced linear system is frequently corrected by the nonlinear term with a specified execution trigger. As a result, computation performance is increased significantly, maintaining sufficient accuracy, which prospectively enables high-performance approximation of nonlinear system behavior.
Keywords
- large-scale systems, Model/controller reduction, modeling
ASJC Scopus subject areas
- Engineering(all)
- Control and Systems Engineering
- Mathematics(all)
- Control and Optimization
Cite this
- Standard
- Harvard
- Apa
- Vancouver
- BibTeX
- RIS
In: IEEE Control Systems Letters, Vol. 2, No. 3, 07.2018, p. 351-356.
Research output: Contribution to journal › Article › Research › peer review
}
TY - JOUR
T1 - Computation-Efficient Simulation of Nonlinear Thermal Boundary Conditions for Large-Scale Models
AU - Frank, Tobias
AU - Bosselmann, Steffen
AU - Wielitzka, Mark
AU - Ortmaier, Tobias
PY - 2018/7
Y1 - 2018/7
N2 - Nonlinear thermal simulations of distributed parameter systems with complex geometry can be performed using finite element analysis (FEA). In order to achieve accurate results, fine spatial and time discretization is required, which often leads to large computation times. However, many methods from system theory, such as parameter identification, real-time model-based control, and model-in-the-loop simulation, heavily rely on either multiple iterations or computation time limits. Hence, a direct model deviation from FEA is unfeasible for these approaches. Model order reduction (MOR) techniques have been proposed to improve computational performance. However, most of them are only applicable to linear systems, but linearization of nonlinear boundary conditions over a wide temperature range does not always fulfill accuracy requirements. Therefore, we propose a simplified nonlinear system description by decoupling nonlinear affected states, performing MOR of the remaining linear term and apply calculated projection to the nonlinear affected part. During simulation, the reduced linear system is frequently corrected by the nonlinear term with a specified execution trigger. As a result, computation performance is increased significantly, maintaining sufficient accuracy, which prospectively enables high-performance approximation of nonlinear system behavior.
AB - Nonlinear thermal simulations of distributed parameter systems with complex geometry can be performed using finite element analysis (FEA). In order to achieve accurate results, fine spatial and time discretization is required, which often leads to large computation times. However, many methods from system theory, such as parameter identification, real-time model-based control, and model-in-the-loop simulation, heavily rely on either multiple iterations or computation time limits. Hence, a direct model deviation from FEA is unfeasible for these approaches. Model order reduction (MOR) techniques have been proposed to improve computational performance. However, most of them are only applicable to linear systems, but linearization of nonlinear boundary conditions over a wide temperature range does not always fulfill accuracy requirements. Therefore, we propose a simplified nonlinear system description by decoupling nonlinear affected states, performing MOR of the remaining linear term and apply calculated projection to the nonlinear affected part. During simulation, the reduced linear system is frequently corrected by the nonlinear term with a specified execution trigger. As a result, computation performance is increased significantly, maintaining sufficient accuracy, which prospectively enables high-performance approximation of nonlinear system behavior.
KW - large-scale systems
KW - Model/controller reduction
KW - modeling
UR - http://www.scopus.com/inward/record.url?scp=85057646622&partnerID=8YFLogxK
U2 - 10.1109/lcsys.2018.2840428
DO - 10.1109/lcsys.2018.2840428
M3 - Article
AN - SCOPUS:85057646622
VL - 2
SP - 351
EP - 356
JO - IEEE Control Systems Letters
JF - IEEE Control Systems Letters
IS - 3
ER -