Ver­öf­fent­li­chun­gen

2011

Zur regelungsorientierten Ableitung von Takagi-Sugeno-Modellen. Kroll, Andreas in Automatisierungstechnik (2011). 59(12) 705–719.
[ Abstract ] [ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
Die Erstellung und Bewertung von Modellen orientiert sich i. d. R. am Prädiktionsfehler oder an der Approximationsgüte unabhängig vom Einsatzzweck eines Modells. Dies liegt an den üblichen Methoden sowie daran, dass Modellbildung und Reglerentwurf häuffig von unterschiedlichen Spezialisten durchgeführt werden. Da selten ein exaktes Modell erstellt werden kann, sollten die Freiheitsgrade bei der Modellbildung genutzt werden, um ein für die geplante Anwendung optimiertes Modell zu erhalten. Beim modellbasierten Reglerentwurf ist die Prädiktionsgüte eines Modells kein primäres Kriterium; vielmehr sollte das Modell gut für den Reglerentwurf geeignet sein. Diese Einsicht führte im Bereich der linearen Systemidentidentifikation zu neuen Methoden. In diesem Beitrag wird eine Methodik zur Ableitung regelungsorientierter Takagi-Sugeno-Fuzzy-Modelle aus physikalischen Modellen vorgestellt und am Benchmark-Beispiel des inversen Pendes demonstriert.
@article{krollat2011, abstract = {Die Erstellung und Bewertung von Modellen orientiert sich i. d. R. am Prädiktionsfehler oder an der Approximationsgüte unabhängig vom Einsatzzweck eines Modells. Dies liegt an den üblichen Methoden sowie daran, dass Modellbildung und Reglerentwurf häuffig von unterschiedlichen Spezialisten durchgeführt werden. Da selten ein exaktes Modell erstellt werden kann, sollten die Freiheitsgrade bei der Modellbildung genutzt werden, um ein für die geplante Anwendung optimiertes Modell zu erhalten. Beim modellbasierten Reglerentwurf ist die Prädiktionsgüte eines Modells kein primäres Kriterium; vielmehr sollte das Modell gut für den Reglerentwurf geeignet sein. Diese Einsicht führte im Bereich der linearen Systemidentidentifikation zu neuen Methoden. In diesem Beitrag wird eine Methodik zur Ableitung regelungsorientierter Takagi-Sugeno-Fuzzy-Modelle aus physikalischen Modellen vorgestellt und am Benchmark-Beispiel des inversen Pendes demonstriert.}, author = {Kroll, Andreas}, journal = {Automatisierungstechnik}, keywords = {Takagi-Sugeno-Fuzzy-Systeme isac-www mrt}, month = {December}, number = 12, pages = {705 – 719}, title = {Zur regelungsorientierten Ableitung von Takagi-Sugeno-Modellen}, volume = 59, year = 2011 }
%0 Journal Article %1 krollat2011 %A Kroll, Andreas %D 2011 %J Automatisierungstechnik %N 12 %P 705 -- 719 %R 10.1524/auto.2011.0964 %T Zur regelungsorientierten Ableitung von Takagi-Sugeno-Modellen %U http://www.oldenbourg-link.com/doi/abs/10.1524/auto.2011.0964 %V 59 %X Die Erstellung und Bewertung von Modellen orientiert sich i. d. R. am Prädiktionsfehler oder an der Approximationsgüte unabhängig vom Einsatzzweck eines Modells. Dies liegt an den üblichen Methoden sowie daran, dass Modellbildung und Reglerentwurf häuffig von unterschiedlichen Spezialisten durchgeführt werden. Da selten ein exaktes Modell erstellt werden kann, sollten die Freiheitsgrade bei der Modellbildung genutzt werden, um ein für die geplante Anwendung optimiertes Modell zu erhalten. Beim modellbasierten Reglerentwurf ist die Prädiktionsgüte eines Modells kein primäres Kriterium; vielmehr sollte das Modell gut für den Reglerentwurf geeignet sein. Diese Einsicht führte im Bereich der linearen Systemidentidentifikation zu neuen Methoden. In diesem Beitrag wird eine Methodik zur Ableitung regelungsorientierter Takagi-Sugeno-Fuzzy-Modelle aus physikalischen Modellen vorgestellt und am Benchmark-Beispiel des inversen Pendes demonstriert.

2019

A Data Selection Method for large Databases for System Identification of MISO Models Based on Recursive Instrumental Variables. Arengas, David; Kroll, Andreas (2019). 357–362.
[ Abstract ] [ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
Experiments to collect data for system identification in industrial plants are constrained due to production and safety requirements. In such situations, logged historical data can be used for system identification instead. However, these recorded data are predominantly stationary in continuously operated plants since processes are seldom excited during normal operation. Performing system identification with such data will yield numerical problems. Alternately, the "most" informative sequences can be extracted and used for system identification. Current data selection methods have several drawbacks. They are constrained to Single-Input Single-Output (SISO) modeling problems. The methods are not robust against correlated noise which is a disadvantage when using real data sets. Moreover, setting design parameters requires some information about the process that is not always available. In this contribution, an alternative data selection method for system identification is presented and evaluated in a case study. In contrast to current approaches, the proposed method does not require data normalization to detect transient changes. It can be used in Multi-Input Single-Output (MISO) systems operating in open or closed loop. An instrumental variables (IV) method is used in the algorithm which provides robustness against non-white noise. Results from a simulation case study of a multivariable system show that models with similar accuracy are obtained when using the intervals retrieved by the data selection method as when using the entire data set.
@inproceedings{arengasecc2019, abstract = {Experiments to collect data for system identification in industrial plants are constrained due to production and safety requirements. In such situations, logged historical data can be used for system identification instead. However, these recorded data are predominantly stationary in continuously operated plants since processes are seldom excited during normal operation. Performing system identification with such data will yield numerical problems. Alternately, the "most" informative sequences can be extracted and used for system identification. Current data selection methods have several drawbacks. They are constrained to Single-Input Single-Output (SISO) modeling problems. The methods are not robust against correlated noise which is a disadvantage when using real data sets. Moreover, setting design parameters requires some information about the process that is not always available. In this contribution, an alternative data selection method for system identification is presented and evaluated in a case study. In contrast to current approaches, the proposed method does not require data normalization to detect transient changes. It can be used in Multi-Input Single-Output (MISO) systems operating in open or closed loop. An instrumental variables (IV) method is used in the algorithm which provides robustness against non-white noise. Results from a simulation case study of a multivariable system show that models with similar accuracy are obtained when using the intervals retrieved by the data selection method as when using the entire data set.}, address = {Naples, Italy}, author = {Arengas, David and Kroll, Andreas}, booktitle = {Proceedings of the 2019 European Control Conference (ECC)}, keywords = {Time-series isac-www mrt}, month = {25.-28. Juni}, organization = {IFAC}, pages = {357-362}, title = {A Data Selection Method for large Databases for System Identification of MISO Models Based on Recursive Instrumental Variables}, year = 2019 }
%0 Conference Paper %1 arengasecc2019 %A Arengas, David %A Kroll, Andreas %B Proceedings of the 2019 European Control Conference (ECC) %C Naples, Italy %D 2019 %P 357-362 %R 10.23919/ECC.2019.8796086 %T A Data Selection Method for large Databases for System Identification of MISO Models Based on Recursive Instrumental Variables %U https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=87960862019 %X Experiments to collect data for system identification in industrial plants are constrained due to production and safety requirements. In such situations, logged historical data can be used for system identification instead. However, these recorded data are predominantly stationary in continuously operated plants since processes are seldom excited during normal operation. Performing system identification with such data will yield numerical problems. Alternately, the "most" informative sequences can be extracted and used for system identification. Current data selection methods have several drawbacks. They are constrained to Single-Input Single-Output (SISO) modeling problems. The methods are not robust against correlated noise which is a disadvantage when using real data sets. Moreover, setting design parameters requires some information about the process that is not always available. In this contribution, an alternative data selection method for system identification is presented and evaluated in a case study. In contrast to current approaches, the proposed method does not require data normalization to detect transient changes. It can be used in Multi-Input Single-Output (MISO) systems operating in open or closed loop. An instrumental variables (IV) method is used in the algorithm which provides robustness against non-white noise. Results from a simulation case study of a multivariable system show that models with similar accuracy are obtained when using the intervals retrieved by the data selection method as when using the entire data set.
Zur Schätzung zulässiger Parametermengen nichtlinearer Takagi-Sugeno-Multi-Modelle mit garantierten Fehlerschranken. Wittich, Felix; Kahl, Matthias; Kroll, Andreas (2019). 247–254.
[ BibTeX ] [ EndNote ] [ DOI ]
 
@inproceedings{wittichgma2019, address = {Dortmund}, author = {Wittich, Felix and Kahl, Matthias and Kroll, Andreas}, booktitle = {29. Workshop Computational Intelligence}, keywords = {isac-www mrt}, month = {28.-29. November}, organization = {GMA-FA 5.14}, pages = {247-254}, publisher = {KIT Scientific Publishing}, title = {Zur Schätzung zulässiger Parametermengen nichtlinearer Takagi-Sugeno-Multi-Modelle mit garantierten Fehlerschranken}, year = 2019 }
%0 Conference Paper %1 wittichgma2019 %A Wittich, Felix %A Kahl, Matthias %A Kroll, Andreas %B 29. Workshop Computational Intelligence %C Dortmund %D 2019 %I KIT Scientific Publishing %P 247-254 %R 10.5445/KSP/1000098736 %T Zur Schätzung zulässiger Parametermengen nichtlinearer Takagi-Sugeno-Multi-Modelle mit garantierten Fehlerschranken
On Using Gated Recurrent Units for Nonlinear System Identification. Rehmer, Alexander; Kroll, Andreas (2019). 2504–2509.
[ Abstract ] [ BibTeX ] [ EndNote ] [ URL ]
 
This paper is concerned with the test signal design for the identification of the partition parameters of locally affine Takagi-Sugeno-(TS-)Models. The basic idea is that data should be generated in local model transition areas in the scheduling space. A reference system output that represents the desired path in the scheduling space is forced upon the underlying system through a combination of a feed-forward and feedback control scheme. The system is then identified in an iterative closed-loop identification procedure. This method has been applied to an artificial system to demonstrate its potential.
@inproceedings{rehmerecc2019, abstract = {This paper is concerned with the test signal design for the identification of the partition parameters of locally affine Takagi-Sugeno-(TS-)Models. The basic idea is that data should be generated in local model transition areas in the scheduling space. A reference system output that represents the desired path in the scheduling space is forced upon the underlying system through a combination of a feed-forward and feedback control scheme. The system is then identified in an iterative closed-loop identification procedure. This method has been applied to an artificial system to demonstrate its potential.}, address = {Naples, Italy}, author = {Rehmer, Alexander and Kroll, Andreas}, booktitle = {Preprints of the 18th European Control Conference (ECC)}, keywords = {isac-www mrt}, month = {25.-28. Juni}, organization = {IFAC}, pages = {2504-2509}, title = {On Using Gated Recurrent Units for Nonlinear System Identification}, year = 2019 }
%0 Conference Paper %1 rehmerecc2019 %A Rehmer, Alexander %A Kroll, Andreas %B Preprints of the 18th European Control Conference (ECC) %C Naples, Italy %D 2019 %P 2504-2509 %T On Using Gated Recurrent Units for Nonlinear System Identification %U https://www.ifac-control.org/events/european-control-conference-in-cooperation-with-ifac-ecc-2019 %X This paper is concerned with the test signal design for the identification of the partition parameters of locally affine Takagi-Sugeno-(TS-)Models. The basic idea is that data should be generated in local model transition areas in the scheduling space. A reference system output that represents the desired path in the scheduling space is forced upon the underlying system through a combination of a feed-forward and feedback control scheme. The system is then identified in an iterative closed-loop identification procedure. This method has been applied to an artificial system to demonstrate its potential.
Zur Homogenisierung von Testsignalen für die nichtlineare Systemidentifikation. Gringard, Matthias; Kroll, Andreas in at -- Automatisierungstechnik (2019). 67(10) 820–832.
[ BibTeX ] [ EndNote ]
 
@article{gringardat2019, author = {Gringard, Matthias and Kroll, Andreas}, journal = {at – Automatisierungstechnik}, keywords = {isac-www mrt}, number = 10, pages = {820–832}, title = {Zur Homogenisierung von Testsignalen für die nichtlineare Systemidentifikation}, volume = 67, year = 2019 }
%0 Journal Article %1 gringardat2019 %A Gringard, Matthias %A Kroll, Andreas %D 2019 %J at -- Automatisierungstechnik %N 10 %P 820--832 %T Zur Homogenisierung von Testsignalen für die nichtlineare Systemidentifikation %V 67
On Nonlinear Empirical Modeling of Residual Stress Profiles in Hard Turning. Wittich, Felix; Kahl, Matthias; Kroll, Andreas; Zinn, Wolfgang; Niendorf, Thomas (2019). 3235–3240.
[ BibTeX ] [ EndNote ]
 
@inproceedings{wittichsmc2019, address = {Bari, Italy}, author = {Wittich, Felix and Kahl, Matthias and Kroll, Andreas and Zinn, Wolfgang and Niendorf, Thomas}, booktitle = {IEEE International Conference on Systems, Man, and Cybernetics (SMC 2019)}, keywords = {isac-www mrt}, month = {06.-09. October}, organization = {IEEE}, pages = {3235 – 3240}, title = {On Nonlinear Empirical Modeling of Residual Stress Profiles in Hard Turning}, year = 2019 }
%0 Conference Paper %1 wittichsmc2019 %A Wittich, Felix %A Kahl, Matthias %A Kroll, Andreas %A Zinn, Wolfgang %A Niendorf, Thomas %B IEEE International Conference on Systems, Man, and Cybernetics (SMC 2019) %C Bari, Italy %D 2019 %P 3235 -- 3240 %T On Nonlinear Empirical Modeling of Residual Stress Profiles in Hard Turning

2020

On data-driven nonlinear uncertainty modeling: Methods and application for control-oriented surface condition prediction in hard turning. Wittich, Felix; Kistner, Lars; Kroll, Andreas; Schott, Christopher; Niendorf, Thomas in tm -- Technisches Messen (2020). 87 732–741.
[ Abstract ] [ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
In this article, two data-driven modeling approaches are investigated, which allow an explicit modeling of uncertainty. For this purpose, parametric Takagi-Sugeno multi-models with bounded-error parameter estimation and nonparametric Gaussian process regression are applied and compared. These models can for instance be used for robust model-based control design. As an application, the prediction of residual stresses during hard turning depending on the machining parameters and the initial hardness is considered.
@article{2020-fw-tm-be_gpr_hardturning, abstract = {In this article, two data-driven modeling approaches are investigated, which allow an explicit modeling of uncertainty. For this purpose, parametric Takagi-Sugeno multi-models with bounded-error parameter estimation and nonparametric Gaussian process regression are applied and compared. These models can for instance be used for robust model-based control design. As an application, the prediction of residual stresses during hard turning depending on the machining parameters and the initial hardness is considered.}, author = {Wittich, Felix and Kistner, Lars and Kroll, Andreas and Schott, Christopher and Niendorf, Thomas}, journal = {tm – Technisches Messen}, keywords = {isac-www mrt}, pages = {732-741}, title = {On data-driven nonlinear uncertainty modeling: Methods and application for control-oriented surface condition prediction in hard turning}, volume = 87, year = 2020 }
%0 Journal Article %1 2020-fw-tm-be_gpr_hardturning %A Wittich, Felix %A Kistner, Lars %A Kroll, Andreas %A Schott, Christopher %A Niendorf, Thomas %D 2020 %J tm -- Technisches Messen %P 732-741 %R 10.1515/teme-2020-0057 %T On data-driven nonlinear uncertainty modeling: Methods and application for control-oriented surface condition prediction in hard turning %U https://doi.org/10.1515/teme-2020-0057 %V 87 %X In this article, two data-driven modeling approaches are investigated, which allow an explicit modeling of uncertainty. For this purpose, parametric Takagi-Sugeno multi-models with bounded-error parameter estimation and nonparametric Gaussian process regression are applied and compared. These models can for instance be used for robust model-based control design. As an application, the prediction of residual stresses during hard turning depending on the machining parameters and the initial hardness is considered.
Toolbox zum Testsignalentwurf für Standardtestsignale für die Identifikation von Eingrößensystemen: Prozessmodellfreie und -basierte Methoden. Himmelsbach, Matthias; Kroll, Andreas (2020).
[ BibTeX ] [ EndNote ] [ URL ]
 
@inproceedings{himmelsbachgmaci2020, address = {Berlin}, author = {Himmelsbach, Matthias and Kroll, Andreas}, booktitle = {30. Workshop Computational Intelligence}, keywords = {isac-www mrt}, month = {26.-27. November}, organization = {GMA-FA 5.14}, publisher = {KIT Scientific Publishing}, title = {Toolbox zum Testsignalentwurf für Standardtestsignale für die Identifikation von Eingrößensystemen: Prozessmodellfreie und -basierte Methoden}, year = 2020 }
%0 Conference Paper %1 himmelsbachgmaci2020 %A Himmelsbach, Matthias %A Kroll, Andreas %B 30. Workshop Computational Intelligence %C Berlin %D 2020 %I KIT Scientific Publishing %T Toolbox zum Testsignalentwurf für Standardtestsignale für die Identifikation von Eingrößensystemen: Prozessmodellfreie und -basierte Methoden %U http://www.rst.e-technik.tu-dortmund.de/cms/de/Veranstaltungen/GMA-Fachausschuss/index.html
Einsatz von maschinellem Lernen für die Rezyklat-Verarbeitung. Moritzer, Elmar; Hopp, Matthias; Wittke, Marius; Deuse, Jochen; Richter, Ralph; Schmitt, Jacqueline; Kroll, Lukas Schulte Andreas; Schrodt, Alexander in WAK -- Jahresmagazin Ingenieurwissenschaften (2020). 70–73.
[ Abstract ] [ BibTeX ] [ EndNote ] [ URL ]
 
Mit Hilfe von ML soll eine Echtzeitregelung in die Lage versetzt werden, nicht sortenreine Rezyklate wie Neugranulat spritzzugießen
@article{2020-kroll_schrodt-wak-rezyklat_ml, abstract = {Mit Hilfe von ML soll eine Echtzeitregelung in die Lage versetzt werden, nicht sortenreine Rezyklate wie Neugranulat spritzzugießen}, author = {Moritzer, Elmar and Hopp, Matthias and Wittke, Marius and Deuse, Jochen and Richter, Ralph and Schmitt, Jacqueline and Kroll, Lukas Schulte Andreas and Schrodt, Alexander}, journal = {WAK – Jahresmagazin Ingenieurwissenschaften}, keywords = {isac-www mrt}, pages = {70 – 73}, title = {Einsatz von maschinellem Lernen für die Rezyklat-Verarbeitung}, year = 2020 }
%0 Journal Article %1 2020-kroll_schrodt-wak-rezyklat_ml %A Moritzer, Elmar %A Hopp, Matthias %A Wittke, Marius %A Deuse, Jochen %A Richter, Ralph %A Schmitt, Jacqueline %A Kroll, Lukas Schulte Andreas %A Schrodt, Alexander %D 2020 %J WAK -- Jahresmagazin Ingenieurwissenschaften %P 70 -- 73 %T Einsatz von maschinellem Lernen für die Rezyklat-Verarbeitung %U http://institut-wv.de/jahresmagazin-ingenieurwissenschaften-im-fokus-kunststofftechnik/ %X Mit Hilfe von ML soll eine Echtzeitregelung in die Lage versetzt werden, nicht sortenreine Rezyklate wie Neugranulat spritzzugießen
Data Selection for System Identification (DS4SID) from Logged Process Records of Continuously Operated Plants. Arengas, David; Kroll, Andreas in at -- Automatisierungstechnik (2020). 68(5) 347–359.
[ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
@article{arengas_at2020, author = {Arengas, David and Kroll, Andreas}, journal = {at – Automatisierungstechnik}, keywords = {isac-www mrt}, number = 5, pages = {347–359}, title = {Data Selection for System Identification (DS4SID) from Logged Process Records of Continuously Operated Plants}, volume = 68, year = 2020 }
%0 Journal Article %1 arengas_at2020 %A Arengas, David %A Kroll, Andreas %D 2020 %J at -- Automatisierungstechnik %N 5 %P 347--359 %R 10.1515/auto-2019-0055 %T Data Selection for System Identification (DS4SID) from Logged Process Records of Continuously Operated Plants %U https://www.degruyter.com/view/journals/auto/68/5/article-p347.xml %V 68
On Scene Flow Computation of Gas Structures with Optical Gas Imaging Cameras. Rangel, J.; Schmoll, R.; Kroll, A. (2020). 174–182.
[ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
@inproceedings{rangel2020, address = {Snowmass Village, Colorado, USA}, author = {Rangel, J. and Schmoll, R. and Kroll, A.}, booktitle = {IEEE Winter Conference on Applications of Computer Vision (WACV 2020)}, keywords = {isac-www mrt}, month = {2.-5. März}, pages = {174-182}, title = {On Scene Flow Computation of Gas Structures with Optical Gas Imaging Cameras}, year = 2020 }
%0 Conference Paper %1 rangel2020 %A Rangel, J. %A Schmoll, R. %A Kroll, A. %B IEEE Winter Conference on Applications of Computer Vision (WACV 2020) %C Snowmass Village, Colorado, USA %D 2020 %P 174-182 %R 10.1109/WACV45572.2020.9093630 %T On Scene Flow Computation of Gas Structures with Optical Gas Imaging Cameras %U https://ieeexplore.ieee.org/document/9093630 %@ 978-1-7281-6554-7
On the vanishing and exploding gradient problem in Gated Recurrent Units. Rehmer, Alexander; Kroll, Andreas (2020).
[ BibTeX ] [ EndNote ]
 
@inproceedings{rehmer-ifac-2020, address = {Berlin, Germany}, author = {Rehmer, Alexander and Kroll, Andreas}, booktitle = {21st IFAC World Congress}, keywords = {isac-www mrt}, month = {July 12-17}, title = {On the vanishing and exploding gradient problem in Gated Recurrent Units}, year = 2020 }
%0 Conference Paper %1 rehmer-ifac-2020 %A Rehmer, Alexander %A Kroll, Andreas %B 21st IFAC World Congress %C Berlin, Germany %D 2020 %T On the vanishing and exploding gradient problem in Gated Recurrent Units
Extending Regularized Least Squares Support Vector Machines for Order Selection of Dynamical Takagi-Sugeno Models. Kahl, Matthias; Kroll, Andreas (2020). (Vol. 53) 1182–1187.
[ Abstract ] [ BibTeX ] [ EndNote ]
 
In this paper, the problem of order selection for nonlinear dynamical Takagi-Sugeno (TS) fuzzy models is adressed. It is solved by reformulating the TS model in its Linear Parameter Varying (LPV) form and applying an extension of a recently proposed Regularized Least Squares Support Vector Machine (R-LSSVM) technique for LPV models. For that, a nonparametric formulation of the TS identi cation problem is proposed which uses data-dependent basis functions. By doing so, the partition of unity of the TS model is preserved and the scheduling dependencies of the model are obtained in a nonparametric manner. For the local order selection, a regularization approach is used which forces the coeffcient functions of insignifcant values of the lagged input and output towards zero.
@inproceedings{kahl-ifac-2020, abstract = {In this paper, the problem of order selection for nonlinear dynamical Takagi-Sugeno (TS) fuzzy models is adressed. It is solved by reformulating the TS model in its Linear Parameter Varying (LPV) form and applying an extension of a recently proposed Regularized Least Squares Support Vector Machine (R-LSSVM) technique for LPV models. For that, a nonparametric formulation of the TS identi cation problem is proposed which uses data-dependent basis functions. By doing so, the partition of unity of the TS model is preserved and the scheduling dependencies of the model are obtained in a nonparametric manner. For the local order selection, a regularization approach is used which forces the coeffcient functions of insignifcant values of the lagged input and output towards zero.}, author = {Kahl, Matthias and Kroll, Andreas}, journal = {IFAC-PapersOnLine}, keywords = {isac-www mrt}, number = 2, pages = {1182–1187}, publisher = {Elsevier}, title = {Extending Regularized Least Squares Support Vector Machines for Order Selection of Dynamical Takagi-Sugeno Models}, volume = 53, year = 2020 }
%0 Conference Paper %1 kahl-ifac-2020 %A Kahl, Matthias %A Kroll, Andreas %D 2020 %I Elsevier %J IFAC-PapersOnLine %N 2 %P 1182--1187 %T Extending Regularized Least Squares Support Vector Machines for Order Selection of Dynamical Takagi-Sugeno Models %V 53 %X In this paper, the problem of order selection for nonlinear dynamical Takagi-Sugeno (TS) fuzzy models is adressed. It is solved by reformulating the TS model in its Linear Parameter Varying (LPV) form and applying an extension of a recently proposed Regularized Least Squares Support Vector Machine (R-LSSVM) technique for LPV models. For that, a nonparametric formulation of the TS identi cation problem is proposed which uses data-dependent basis functions. By doing so, the partition of unity of the TS model is preserved and the scheduling dependencies of the model are obtained in a nonparametric manner. For the local order selection, a regularization approach is used which forces the coeffcient functions of insignifcant values of the lagged input and output towards zero.
Evaluierung von Verfahren zur optischen Bestimmung von Gasgeschwindigkeiten. Dierks, Sören; Kroll, Andreas in TM -- Technisches Messen (2020). 87(1) 66–77.
[ Abstract ] [ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
In diesem Beitrag werden verschiedene Verfahren zur Bestimmung des optischen Flusses auf ihre Eignung zur Bestimmung von Gasgeschwindigkeiten hin untersucht. Dazu werden die Verfahren auf Differenzbilder einer sogenannten Gaskamera angewandt, um eine mittlere Gasgeschwindigkeit zu bestimmen. Mit einem Prüfstand im Laborumfeld werden die Verfahrencharakterisiert und abschließend wird die Messunsicherheita bgeschätzt.
@article{2020-sd-tm-eval_gasgeschwindigkeit, abstract = {In diesem Beitrag werden verschiedene Verfahren zur Bestimmung des optischen Flusses auf ihre Eignung zur Bestimmung von Gasgeschwindigkeiten hin untersucht. Dazu werden die Verfahren auf Differenzbilder einer sogenannten Gaskamera angewandt, um eine mittlere Gasgeschwindigkeit zu bestimmen. Mit einem Prüfstand im Laborumfeld werden die Verfahrencharakterisiert und abschließend wird die Messunsicherheita bgeschätzt.}, author = {Dierks, Sören and Kroll, Andreas}, journal = {TM – Technisches Messen}, keywords = {isac-www mrt}, number = 1, pages = {66–77}, title = {Evaluierung von Verfahren zur optischen Bestimmung von Gasgeschwindigkeiten}, volume = 87, year = 2020 }
%0 Journal Article %1 2020-sd-tm-eval_gasgeschwindigkeit %A Dierks, Sören %A Kroll, Andreas %D 2020 %J TM -- Technisches Messen %N 1 %P 66--77 %R 10.1515/teme-2019-0078 %T Evaluierung von Verfahren zur optischen Bestimmung von Gasgeschwindigkeiten %U https://www.degruyter.com/view/j/teme.ahead-of-print/teme-2019-0078/teme-2019-0078.xml %V 87 %X In diesem Beitrag werden verschiedene Verfahren zur Bestimmung des optischen Flusses auf ihre Eignung zur Bestimmung von Gasgeschwindigkeiten hin untersucht. Dazu werden die Verfahren auf Differenzbilder einer sogenannten Gaskamera angewandt, um eine mittlere Gasgeschwindigkeit zu bestimmen. Mit einem Prüfstand im Laborumfeld werden die Verfahrencharakterisiert und abschließend wird die Messunsicherheita bgeschätzt.
On considering the output in space-filling test signal designs for the identification of dynamic Takagi-Sugeno models. Gringard, Matthias; Kroll, Andreas (2020).
[ Abstract ] [ BibTeX ] [ EndNote ] [ URL ]
 
In this contribution the optimal experiment design for the identification of linear affine Takagi-Sugeno (TS) models is discussed. The parameters of these models can be separated into local model and partition parameters which also allows for a separation of the design. The presented optimal design is based on the Fisher Information Matrix (FIM) and its objective is to minimize the parameter estimation uncertainty. In this contribution parameters of a standard test signal type (multi-step) are optimized to achieve this objective. The effectivity of FIM-based designs for nonlinear models depends on initial identification including structural decisions. The general functionality of the presented method is demonstrated on a case study.
@inproceedings{gringardifac2020, abstract = {In this contribution the optimal experiment design for the identification of linear affine Takagi-Sugeno (TS) models is discussed. The parameters of these models can be separated into local model and partition parameters which also allows for a separation of the design. The presented optimal design is based on the Fisher Information Matrix (FIM) and its objective is to minimize the parameter estimation uncertainty. In this contribution parameters of a standard test signal type (multi-step) are optimized to achieve this objective. The effectivity of FIM-based designs for nonlinear models depends on initial identification including structural decisions. The general functionality of the presented method is demonstrated on a case study.}, address = {Berlin, Germany}, author = {Gringard, Matthias and Kroll, Andreas}, booktitle = {21st IFAC World Congress}, keywords = {isac-www mrt}, month = {12-17. July}, organization = {IFAC}, title = {On considering the output in space-filling test signal designs for the identification of dynamic {T}akagi-{S}ugeno models}, year = 2020 }
%0 Conference Paper %1 gringardifac2020 %A Gringard, Matthias %A Kroll, Andreas %B 21st IFAC World Congress %C Berlin, Germany %D 2020 %T On considering the output in space-filling test signal designs for the identification of dynamic Takagi-Sugeno models %U https://www.ifac2020.org/ %X In this contribution the optimal experiment design for the identification of linear affine Takagi-Sugeno (TS) models is discussed. The parameters of these models can be separated into local model and partition parameters which also allows for a separation of the design. The presented optimal design is based on the Fisher Information Matrix (FIM) and its objective is to minimize the parameter estimation uncertainty. In this contribution parameters of a standard test signal type (multi-step) are optimized to achieve this objective. The effectivity of FIM-based designs for nonlinear models depends on initial identification including structural decisions. The general functionality of the presented method is demonstrated on a case study.

2021

Prediction of near surface residual stress states for hard turned specimens using data driven nonlinear models. Schott, Christopher; Wittich, Felix; Kroll, Andreas; Niendorf, Thomas (2021).
[ BibTeX ] [ EndNote ]
 
@inproceedings{schott_cirp2020, address = {Sheffield, UK}, author = {Schott, Christopher and Wittich, Felix and Kroll, Andreas and Niendorf, Thomas}, booktitle = {9th CIRP Conference on High Performance Cutting}, keywords = {isac-www mrt}, note = {{accepted}}, title = {Prediction of near surface residual stress states for hard turned specimens using data driven nonlinear models}, year = 2021 }
%0 Conference Paper %1 schott_cirp2020 %A Schott, Christopher %A Wittich, Felix %A Kroll, Andreas %A Niendorf, Thomas %B 9th CIRP Conference on High Performance Cutting %C Sheffield, UK %D 2021 %T Prediction of near surface residual stress states for hard turned specimens using data driven nonlinear models
Multispectral Geometric Calibration of Cameras in Visual and Infrared Spectral Range. Schramm, Sebastian; Rangel, Johannes; Aguirre Salazar, Daniela; Schmoll, Robert; Kroll, Andreas in IEEE Sensors (2021). 21(2) 2159–2168.
[ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
@article{schrammrangel2021, author = {Schramm, Sebastian and Rangel, Johannes and Aguirre Salazar, Daniela and Schmoll, Robert and Kroll, Andreas}, journal = {IEEE Sensors}, keywords = {isac-www mrt}, number = 2, pages = {2159-2168}, title = {Multispectral Geometric Calibration of Cameras in Visual and Infrared Spectral Range}, volume = 21, year = 2021 }
%0 Journal Article %1 schrammrangel2021 %A Schramm, Sebastian %A Rangel, Johannes %A Aguirre Salazar, Daniela %A Schmoll, Robert %A Kroll, Andreas %D 2021 %J IEEE Sensors %N 2 %P 2159-2168 %R 10.1109/JSEN.2020.3019959 %T Multispectral Geometric Calibration of Cameras in Visual and Infrared Spectral Range %U https://ieeexplore.ieee.org/document/9178752 %V 21
Catadioptric Stereo Optical Gas Imaging System for Scene Flow Computation of Gas Structures. Rangel, J.; Schmoll, R.; Kroll, A. in IEEE Sensors Journal (2021). 21(5) 6811–6820.
[ BibTeX ] [ EndNote ] [ URL ] [ DOI ]
 
@article{rangel_ieeesensors_2020, author = {Rangel, J. and Schmoll, R. and Kroll, A.}, journal = {IEEE Sensors Journal}, keywords = {isac-www mrt}, number = 5, pages = {6811 – 6820}, title = {Catadioptric Stereo Optical Gas Imaging System for Scene Flow Computation of Gas Structures}, volume = 21, year = 2021 }
%0 Journal Article %1 rangel_ieeesensors_2020 %A Rangel, J. %A Schmoll, R. %A Kroll, A. %D 2021 %J IEEE Sensors Journal %N 5 %P 6811 -- 6820 %R https://doi.org/10.1109/JSEN.2020.3042116 %T Catadioptric Stereo Optical Gas Imaging System for Scene Flow Computation of Gas Structures %U https://ieeexplore.ieee.org/document/9277592 %V 21
3D Thermography for the Measurement of Surface Heat Dissipation. Schmoll, Robert; Schramm, Sebastian; Breitenstein, Tom; Kroll, Andreas (2021).
[ BibTeX ] [ EndNote ]
 
@inproceedings{schmollsmsi2021, address = {Nuremberg, Germany}, author = {Schmoll, Robert and Schramm, Sebastian and Breitenstein, Tom and Kroll, Andreas}, booktitle = {Sensor and Measurement Science International Conference (SMSI 2021)}, keywords = {isac-www mrt}, month = {03.-06. May}, note = {{accepted}}, organization = {AMA Verband für Sensorik und Messtechnik e.V.}, title = {3D Thermography for the Measurement of Surface Heat Dissipation}, year = 2021 }
%0 Conference Paper %1 schmollsmsi2021 %A Schmoll, Robert %A Schramm, Sebastian %A Breitenstein, Tom %A Kroll, Andreas %B Sensor and Measurement Science International Conference (SMSI 2021) %C Nuremberg, Germany %D 2021 %T 3D Thermography for the Measurement of Surface Heat Dissipation
Zur approximativen Maximum-Likelihood-Schätzung dynamischer Multi-Modelle vom Typ Takagi-Sugeno: Methodik und Anwendung auf einen Servo-Pneumatikantrieb. Kroll, Andreas; Fischer, Jana in at -- Automatisierungstechnik (2021).
[ Abstract ] [ BibTeX ] [ EndNote ]
 
Ist Wissen über die stochastischen Eigenschaften eines dynamischen Prozesses verfügbar, so kann dies zur Verbesserung der Modellparameterschätzung berücksichtigt werden. Dabei trifft die oft gemachte Annahme i.i.d. normalverteilter Zufallsgrößen bei technischen Systemen häufig nicht zu. Im Beitrag werden deshalb die Dichtefunktionen der Zufallsgrößen als Gaußsche Mischmodelle mittels eines Multistart-Expectation-Maximization-Algorithmus geschätzt. Da die Residualdichtefunktionen zu Beginn der Identifikation nicht verfügbar sind, werden diese aus der Streuung von Wiederholexperimenten approximativ ermittelt und zur Maximum-Likelihood-Schätzung von Takagi-Sugeno-Modellen genutzt. Wegen der ausgeprägten Multimodalität der resultierenden Likelihood-Funktion wird dazu Partikelschwarm-Optimierung eingesetzt. Die Vorgehensweise wird an einem industriellen servo-pneumatischen Linearantrieb demonstriert.
@article{2021-krool_fischer-at-approx_max_likelihood_ts_modelle, abstract = {Ist Wissen über die stochastischen Eigenschaften eines dynamischen Prozesses verfügbar, so kann dies zur Verbesserung der Modellparameterschätzung berücksichtigt werden. Dabei trifft die oft gemachte Annahme i.i.d. normalverteilter Zufallsgrößen bei technischen Systemen häufig nicht zu. Im Beitrag werden deshalb die Dichtefunktionen der Zufallsgrößen als Gaußsche Mischmodelle mittels eines Multistart-Expectation-Maximization-Algorithmus geschätzt. Da die Residualdichtefunktionen zu Beginn der Identifikation nicht verfügbar sind, werden diese aus der Streuung von Wiederholexperimenten approximativ ermittelt und zur Maximum-Likelihood-Schätzung von Takagi-Sugeno-Modellen genutzt. Wegen der ausgeprägten Multimodalität der resultierenden Likelihood-Funktion wird dazu Partikelschwarm-Optimierung eingesetzt. Die Vorgehensweise wird an einem industriellen servo-pneumatischen Linearantrieb demonstriert.}, author = {Kroll, Andreas and Fischer, Jana}, journal = {at – Automatisierungstechnik}, keywords = {isac-www mrt}, note = {accepted}, title = {Zur approximativen Maximum-Likelihood-Schätzung dynamischer Multi-Modelle vom Typ Takagi-Sugeno: Methodik und Anwendung auf einen Servo-Pneumatikantrieb}, year = 2021 }
%0 Journal Article %1 2021-krool_fischer-at-approx_max_likelihood_ts_modelle %A Kroll, Andreas %A Fischer, Jana %D 2021 %J at -- Automatisierungstechnik %T Zur approximativen Maximum-Likelihood-Schätzung dynamischer Multi-Modelle vom Typ Takagi-Sugeno: Methodik und Anwendung auf einen Servo-Pneumatikantrieb %X Ist Wissen über die stochastischen Eigenschaften eines dynamischen Prozesses verfügbar, so kann dies zur Verbesserung der Modellparameterschätzung berücksichtigt werden. Dabei trifft die oft gemachte Annahme i.i.d. normalverteilter Zufallsgrößen bei technischen Systemen häufig nicht zu. Im Beitrag werden deshalb die Dichtefunktionen der Zufallsgrößen als Gaußsche Mischmodelle mittels eines Multistart-Expectation-Maximization-Algorithmus geschätzt. Da die Residualdichtefunktionen zu Beginn der Identifikation nicht verfügbar sind, werden diese aus der Streuung von Wiederholexperimenten approximativ ermittelt und zur Maximum-Likelihood-Schätzung von Takagi-Sugeno-Modellen genutzt. Wegen der ausgeprägten Multimodalität der resultierenden Likelihood-Funktion wird dazu Partikelschwarm-Optimierung eingesetzt. Die Vorgehensweise wird an einem industriellen servo-pneumatischen Linearantrieb demonstriert.