Anomaly detection in injection molding process data based on unsupervised learning

Schiffers, Reinhard

doi:10.3139/o999.02052018

Cited by 4 publications

(1 citation statement)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Still there are influences on the injection molding process, such as variations in the material properties or environmental conditions, which may negatively affect the quality of the molded parts. To address these issues, the research focus over the last few years has moved to improving control techniques [14,18] and analyzing process data [16,17].…”

Section: Case Studymentioning

confidence: 99%

Providing Meaningful Data Summarizations Using Exemplar-based Clustering in Industry 4.0

Honysz,

Schulze-Struchtrup,

Buschjäger

et al. 2021

Preprint

View full text Add to dashboard Cite

Data summarizations are a valuable tool to derive knowledge from large data streams and have proven their usefulness in a great number of applications. Summaries can be found by optimizing submodular functions. These functions map subsets of data to real values, which indicate their "representativeness" and which should be maximized to find a diverse summary of the underlying data. In this paper, we studied Exemplar-based clustering as a submodular function and provide a GPU algorithm to cope with its high computational complexity. We show, that our GPU implementation provides speedups of up to 72x using singleprecision and up to 452x using half-precision computation compared to conventional CPU algorithms. We also show, that the GPU algorithm not only provides remarkable runtime benefits with workstation-grade GPUs but also with low-power embedded computation units for which speedups of up to 35x are possible. Furthermore, we apply our algorithm to realworld data from injection molding manufacturing processes and discuss how found summaries help with steering this specific process to cut costs and reduce the manufacturing of bad parts. Beyond pure speedup considerations, we show, that our approach can provide summaries within reasonable time frames for this kind of industrial, real-world data.

show abstract

Section: Case Studymentioning

confidence: 99%

Providing Meaningful Data Summarizations Using Exemplar-based Clustering in Industry 4.0

Honysz,

Schulze-Struchtrup,

Buschjäger

et al. 2021

Preprint

View full text Add to dashboard Cite

show abstract

Anomaly Detection for Injection Molding Using Probabilistic Deep Learning

Ketonen

Blech

2021

2021 4th IEEE International Conference on Industrial Cyber-Physical Systems (ICPS)

View full text Add to dashboard Cite

Time series data for process monitoring in injection molding: a quantitative study of the benefits of a high sampling rate

Bogedale

Schrodt

Heim

2023

International Polymer Processing

View full text Add to dashboard Cite

Process monitoring systems are playing an increasingly important role in reducing production capacity losses in injection molding. Process monitoring and optimization systems are mostly based on processing data of injection molding machine control systems. These data consist of scalar data and time series. This paper introduces a novel approach to modelling injection molding processes using only time series data and evaluates the quantitative influences of varying sampling times on calculation of integral values and model quality. On the basis of the first experiment, it is shown that the sampling rates of these time series have a large influence on information which can be derived from this data (e.g. injection work). These findings provide an assessment of whether the effort is justified for the respective requirements on the accuracy of the injection work and other parameters derived from the time series. In the second experiment, a model is presented which uses only the injection flow and injection pressure profile as input and achieves high coefficients of determination for the prediction of the part weight, despite the absence of mold sensor data and scalar data. It is shown that higher sampling rates of time series results in higher prediction quality of these models. This improves the understanding of the data needed for high quality machine learning models of injection molding processes and enable users to estimate a lower bound for the sample rates of time series for their use cases.

show abstract

Anomaly detection in injection molding process data based on unsupervised learning

Cited by 4 publications

References 19 publications

Providing Meaningful Data Summarizations Using Exemplar-based Clustering in Industry 4.0

Providing Meaningful Data Summarizations Using Exemplar-based Clustering in Industry 4.0

Anomaly Detection for Injection Molding Using Probabilistic Deep Learning

Time series data for process monitoring in injection molding: a quantitative study of the benefits of a high sampling rate

Contact Info

Product

Resources

About