CoChat
Sign Up
Cited by 4Fault Detection and Control SystemsSpectroscopy and Chemometric AnalysesFuzzy Logic and Control Systems

Evolving time-series based prediction models for quality criteria in a multi-stage production process

Lughofer, Edwin, Pollak, Robert, Zavoianu, Alexandru-Ciprian, Meyer-Heye, Pauline, Zorrer, Helmut, Eitzinger, Christian, Lehner, Jasmin, Radauer, Thomas +1 more·2018·2018 IEEE Conference on Evolving and Adaptive Intelligent Systems (EAIS)
DOI 10.1109/eais.2018.8397186
Open this paper in CoChat — free account

Abstract

We address the problem of predicting product quality for a latter stage in a production process already at an early stage. Thereby, the idea is to use time-series of process values, recorded during the on-line production process of the early stage and containing possible system dynamics and variations according to parameter settings or different environmental conditions, as input to predict the final quality criteria in the latter stage. We apply a non-linear partial least squares (PLS) variant for reducing the high input dimensionality of time-series batch-process problems, by combining PLS with generalized Takagi-Sugeno fuzzy systems, a new extended variant of classical TS fuzzy system (thus termed as PLS-Fuzzy). This combination opens the possibility to resolve non-linearities in the PLS score space without requiring extra pre-tuning parameters (as is the case in other non-linear PLS variants). The models are trained by an evolving and iterative vector quantization approach to find the optimal number of rules and their ideal positioning and shape, combined with a fuzzily weighted version of elastic net regu-larization for robust estimation of the consequent parameters. The adaptation algorithm of the models (termed as IPLS-GEFS) includes an on-the-fly evolving rule learning concept (GEFS), a forgetting strategy with dynamically varying forgetting factor in case of drifts (to increase flexibility by outweighing older samples) as well as a new variant for an incremental single-pass update of the latent variable space (IPLS). The latter can be seen as a new variant for incremental dimension reduction and subspace update and is necessary when the covariance characteristics between input and output changes. Results on a real-world data set from microfluidic chip production show a comparable performance of PLS-Fuzzy with random forests, extreme learning machines and deep learning with MLP neural networks, achieving low prediction errors (below 10%) with low model complexity. Updating the models with new on-line data — only achievable with our method, as the others are batch off-line methods (with mostly slow re-training phases) — decreased the model errors, at most when including incremental latent variable space update.

Integrity signals

Registry

Crossref · 10.1109/eais.2018.8397186

Cites

39 of 39 references verified0 citation errors · 0 not checkable

Statistics

Recomputing p-values needs the paper’s full text.

Claims the paper makes

This paper’s full text isn’t openly available, so its claims and self-check weren’t run. The registry and citation checks still apply.

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • Partially corrupted data from field devices affects real-time process control and time series analysis used for equipment reliability analysis, predictive maintenance, process optimization, condition monitoring, and quality control.

    “If there is even partially corrupted data coming from field devices, it affects not only real-time process control, but also time series analysis applied in various algorithms used for equipment reliability analysis i.e. predictive maintenance [44], [45], process optimization, condition monitoring [46], quality control [47], [48] and similar.”

    From SCADA Systems With Focus on Continuous Manufacturing and Steel Industry: A Survey on Architectures, Standards, Challenges and Industry 5.0 (2022) · cites this paper as [46]

  • Another approach to corrupted data from field devices is development of ML algorithms detecting and describing patterns that lead to abrupt changes in behavior of sensor signals in specific operating scenarios.

    “Another approach to corrupted data coming from field devices is the development of ML algorithms detecting and describing the patterns that lead to abrupt changes in the behavior of sensor signals in specific operating scenarios [46].”

    From SCADA Systems With Focus on Continuous Manufacturing and Steel Industry: A Survey on Architectures, Standards, Challenges and Industry 5.0 (2022) · cites this paper as [46]

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
Lughofer, E., Pollak, R., Zavoianu, A.-C., Meyer-Heye, P., Zorrer, H., Eitzinger, C., Lehner, J., Radauer, T., & Pratama, M.
Affiliations
  • Johannes Kepler University of Linz (AT)
  • PROFACTOR GmbH (Austria) (AT)
  • Nanyang Technological University (SG)
ORCID
Venue
2018 IEEE Conference on Evolving and Adaptive Intelligent Systems (EAIS)
Publisher
IEEE
Published
2018-05
Volume
171
Pages
1-10
DOI
10.1109/eais.2018.8397186
OpenAlex
W2810385025
Cited by
4
References
39
Citations by year
2022: 2 · 2020: 1 · 2019: 1
FWCI
0.75
Type
conference paper
Language
en
Open access
closed
Locations
IEEE Conference on Evolving and Adaptive Intelligent Systems (EAIS) (publishedVersion)
Funders
Österreichische Forschungsgemeinschaft
Topics
  • Physical Sciences › Engineering › Control and Systems Engineering › Fault Detection and Control Systems
  • Physical Sciences › Chemistry › Analytical Chemistry › Spectroscopy and Chemometric Analyses
  • Physical Sciences › Computer Science › Artificial Intelligence › Fuzzy Logic and Control Systems
Keywords
product quality prediction, multi-stage production processes, online model adaptation, batch processes

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review