Version 1
: Received: 1 June 2022 / Approved: 2 June 2022 / Online: 2 June 2022 (11:08:35 CEST)
Version 2
: Received: 18 August 2022 / Approved: 19 August 2022 / Online: 19 August 2022 (06:03:08 CEST)
How to cite:
Matlala, K.; Mishra, A.K.; Puthal, D. Towards Sustainable Aluminium Processing: Autonomous Quality Control Using Business Analytics. Preprints2022, 2022060033. https://doi.org/10.20944/preprints202206.0033.v2
Matlala, K.; Mishra, A.K.; Puthal, D. Towards Sustainable Aluminium Processing: Autonomous Quality Control Using Business Analytics. Preprints 2022, 2022060033. https://doi.org/10.20944/preprints202206.0033.v2
Matlala, K.; Mishra, A.K.; Puthal, D. Towards Sustainable Aluminium Processing: Autonomous Quality Control Using Business Analytics. Preprints2022, 2022060033. https://doi.org/10.20944/preprints202206.0033.v2
APA Style
Matlala, K., Mishra, A.K., & Puthal, D. (2022). Towards Sustainable Aluminium Processing: Autonomous Quality Control Using Business Analytics. Preprints. https://doi.org/10.20944/preprints202206.0033.v2
Chicago/Turabian Style
Matlala, K., Amit Kumar Mishra and Deepak Puthal. 2022 "Towards Sustainable Aluminium Processing: Autonomous Quality Control Using Business Analytics" Preprints. https://doi.org/10.20944/preprints202206.0033.v2
Abstract
This paper presents work done as part of a transformation effort towards a greener and more sustainable Aluminium manufacturing plant. The effort includes reducing the carbon footprint by minimising waste and increasing operational efficiency. The contribution of this work includes the reduction of waste through the implementation of autonomous, real-time quality measurement and classification at an Aluminium casthouse. Data is collected from the MV20/20 which uses ultrasound pulses to detect molten Aluminium inclusions, which degrade the quality of the metal and cause subsequent metal waste. The sensor measures cleanliness, inclusion counts and distributions from 20 - 160 microns. The contribution of this work is in the development of business analytics to implement condition-based monitoring through anomaly detection, and to classify inclusion types for samples that failed. For anomaly detection, multivariate K-Means and DBSCAN algorithms are compared as they have been proven to work in a wide range of datasets. For classification, a two-stage classifier is implemented. The first stage classifies the success or failure of the sample, while the second stage classifies the inclusion responsible for the failed sample. The algorithms considered include logistic regression, support vector machine, multi-layer perceptron and radial basis function network. The multi-layer perceptron offers the best performance using k-fold cross-validation, and is further tuned using grid search to explore the possibility of an even better performance. The results reveal that the model has achieved a global maximum in performance. Recommendations include the integration of additional sensor systems and the improvements in quality assurance practices.
Keywords
MV20/20; PoDFA; LiMCA; Business Analytics; anomaly detection; statistical process control; K-Means; DBSCAN; multi-layer perceptron; activation fucntion; inclusion; confusion matrix
Subject
Engineering, Industrial and Manufacturing Engineering
Copyright:
This is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Commenter: Kgothatso Matlala
Commenter's Conflict of Interests: Author
Updated abstract
Updated rationale for work
Updated related works
Updated contribution of this work