Parameter learning algorithms for continuous model improvement using operational data

Anders L. Madsen; Nicolaj Søndberg Jeppesen; Frank Jensen; Mohamed S. Sayed; Ulrich Moser; Luis Neto; Joao Reis; Niels Lohse

doi:10.1007/978-3-319-61581-3_11

Parameter learning algorithms for continuous model improvement using operational data

Anders L. Madsen^*, Nicolaj Søndberg Jeppesen, Frank Jensen, Mohamed S. Sayed, Ulrich Moser, Luis Neto, Joao Reis, Niels Lohse

^*Corresponding author for this work

Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review

2 Citations (Scopus)

180 Downloads (Pure)

Abstract

In this paper, we consider the application of object-oriented Bayesian networks to failure diagnostics in manufacturing systems and continuous model improvement based on operational data. The analysis is based on an object-oriented Bayesian network developed for failure diagnostics of a one-dimensional pick-and-place industrial robot developed by IEF-Werner GmbH.We consider four learning algorithms (batch Expectation-Maximization (EM), incremental EM, Online EM and fractional updating) for parameter updating in the object-oriented Bayesian network using a real operational dataset. Also, we evaluate the performance of the considered algorithms on a dataset generated from the model to determine which algorithm is best suited for recovering the underlying generating distribution. The object-oriented Bayesian network has been integrated into both the control software of the robot as well as into a software architecture that supports diagnostic and prognostic capabilities of devices in manufacturing systems. We evaluate the time performance of the architecture to determine the feasibility of online learning from operational data using each of the four algorithms.

Original language	English
Title of host publication	Symbolic and Quantitative Approaches to Reasoning with Uncertainty : 14th European Conference, ECSQARU 2017, Proceedings
Number of pages	10
Publisher	Springer
Publication date	2017
Pages	115-124
ISBN (Print)	978-3-319-61580-6
ISBN (Electronic)	978-3-319-61581-3
DOIs	https://doi.org/10.1007/978-3-319-61581-3_11
Publication status	Published - 2017
Event	14th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty, ECSQARU 2017 - Lugano, Switzerland Duration: 10 Jul 2017 → 14 Jul 2017

Conference

Conference	14th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty, ECSQARU 2017
Country/Territory	Switzerland
City	Lugano
Period	10/07/2017 → 14/07/2017

Series	Lecture Notes In Artificial Intelligence
Number	10369
ISSN	0302-9743

Keywords

Bayesian networks
Parameter update
Practical application

Access to Document

10.1007/978-3-319-61581-3_11

Parameter_Learning_Algorithms_for_Continuous_Model_Improvement_Using_Operational_DataAccepted author manuscript, 462 KBLicence: CC BY-NC-ND 4.0

AUB Link

Search for the material in Aalborg University Library's search engine

Cite this

Madsen, A. L., Jeppesen, N. S., Jensen, F., Sayed, M. S., Moser, U., Neto, L., Reis, J., & Lohse, N. (2017). Parameter learning algorithms for continuous model improvement using operational data. In Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 14th European Conference, ECSQARU 2017, Proceedings (pp. 115-124). Springer. https://doi.org/10.1007/978-3-319-61581-3_11

@inproceedings{81f481f6737c4cb786737dcbdd33df18,

title = "Parameter learning algorithms for continuous model improvement using operational data",

abstract = "In this paper, we consider the application of object-oriented Bayesian networks to failure diagnostics in manufacturing systems and continuous model improvement based on operational data. The analysis is based on an object-oriented Bayesian network developed for failure diagnostics of a one-dimensional pick-and-place industrial robot developed by IEF-Werner GmbH.We consider four learning algorithms (batch Expectation-Maximization (EM), incremental EM, Online EM and fractional updating) for parameter updating in the object-oriented Bayesian network using a real operational dataset. Also, we evaluate the performance of the considered algorithms on a dataset generated from the model to determine which algorithm is best suited for recovering the underlying generating distribution. The object-oriented Bayesian network has been integrated into both the control software of the robot as well as into a software architecture that supports diagnostic and prognostic capabilities of devices in manufacturing systems. We evaluate the time performance of the architecture to determine the feasibility of online learning from operational data using each of the four algorithms.",

keywords = "Bayesian networks, Parameter update, Practical application",

author = "Madsen, {Anders L.} and Jeppesen, {Nicolaj S{\o}ndberg} and Frank Jensen and Sayed, {Mohamed S.} and Ulrich Moser and Luis Neto and Joao Reis and Niels Lohse",

year = "2017",

doi = "10.1007/978-3-319-61581-3_11",

language = "English",

isbn = "978-3-319-61580-6",

series = "Lecture Notes In Artificial Intelligence",

publisher = "Springer",

number = "10369",

pages = "115--124",

booktitle = "Symbolic and Quantitative Approaches to Reasoning with Uncertainty",

address = "Germany",

note = "14th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty, ECSQARU 2017 ; Conference date: 10-07-2017 Through 14-07-2017",

}

Madsen, AL , Jeppesen, NS, Jensen, F, Sayed, MS, Moser, U, Neto, L, Reis, J & Lohse, N 2017, Parameter learning algorithms for continuous model improvement using operational data. in Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 14th European Conference, ECSQARU 2017, Proceedings. Springer, Lecture Notes In Artificial Intelligence, no. 10369, pp. 115-124, 14th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty, ECSQARU 2017, Lugano, Switzerland, 10/07/2017. https://doi.org/10.1007/978-3-319-61581-3_11

Parameter learning algorithms for continuous model improvement using operational data. / Madsen, Anders L.; Jeppesen, Nicolaj Søndberg; Jensen, Frank et al.
Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 14th European Conference, ECSQARU 2017, Proceedings. Springer, 2017. p. 115-124 (Lecture Notes In Artificial Intelligence; No. 10369).

Research output: Contribution to book/anthology/report/conference proceeding › Article in proceeding › Research › peer-review

TY - GEN

T1 - Parameter learning algorithms for continuous model improvement using operational data

AU - Madsen, Anders L.

AU - Jeppesen, Nicolaj Søndberg

AU - Jensen, Frank

AU - Sayed, Mohamed S.

AU - Moser, Ulrich

AU - Neto, Luis

AU - Reis, Joao

AU - Lohse, Niels

PY - 2017

Y1 - 2017

N2 - In this paper, we consider the application of object-oriented Bayesian networks to failure diagnostics in manufacturing systems and continuous model improvement based on operational data. The analysis is based on an object-oriented Bayesian network developed for failure diagnostics of a one-dimensional pick-and-place industrial robot developed by IEF-Werner GmbH.We consider four learning algorithms (batch Expectation-Maximization (EM), incremental EM, Online EM and fractional updating) for parameter updating in the object-oriented Bayesian network using a real operational dataset. Also, we evaluate the performance of the considered algorithms on a dataset generated from the model to determine which algorithm is best suited for recovering the underlying generating distribution. The object-oriented Bayesian network has been integrated into both the control software of the robot as well as into a software architecture that supports diagnostic and prognostic capabilities of devices in manufacturing systems. We evaluate the time performance of the architecture to determine the feasibility of online learning from operational data using each of the four algorithms.

AB - In this paper, we consider the application of object-oriented Bayesian networks to failure diagnostics in manufacturing systems and continuous model improvement based on operational data. The analysis is based on an object-oriented Bayesian network developed for failure diagnostics of a one-dimensional pick-and-place industrial robot developed by IEF-Werner GmbH.We consider four learning algorithms (batch Expectation-Maximization (EM), incremental EM, Online EM and fractional updating) for parameter updating in the object-oriented Bayesian network using a real operational dataset. Also, we evaluate the performance of the considered algorithms on a dataset generated from the model to determine which algorithm is best suited for recovering the underlying generating distribution. The object-oriented Bayesian network has been integrated into both the control software of the robot as well as into a software architecture that supports diagnostic and prognostic capabilities of devices in manufacturing systems. We evaluate the time performance of the architecture to determine the feasibility of online learning from operational data using each of the four algorithms.

KW - Bayesian networks

KW - Parameter update

KW - Practical application

UR - http://www.scopus.com/inward/record.url?scp=85025175994&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-61581-3_11

DO - 10.1007/978-3-319-61581-3_11

M3 - Article in proceeding

AN - SCOPUS:85025175994

SN - 978-3-319-61580-6

T3 - Lecture Notes In Artificial Intelligence

SP - 115

EP - 124

BT - Symbolic and Quantitative Approaches to Reasoning with Uncertainty

PB - Springer

T2 - 14th European Conference on Symbolic and Quantitative Approaches to Reasoning with Uncertainty, ECSQARU 2017

Y2 - 10 July 2017 through 14 July 2017

ER -

Madsen AL , Jeppesen NS, Jensen F, Sayed MS, Moser U, Neto L et al. Parameter learning algorithms for continuous model improvement using operational data. In Symbolic and Quantitative Approaches to Reasoning with Uncertainty: 14th European Conference, ECSQARU 2017, Proceedings. Springer. 2017. p. 115-124. (Lecture Notes In Artificial Intelligence; No. 10369). doi: 10.1007/978-3-319-61581-3_11

Parameter learning algorithms for continuous model improvement using operational data

Abstract

Conference

Keywords

Access to Document

AUB Link

Other files and links

Fingerprint

Cite this