Contextual classification for smart machining based on unsupervised machine learning by Gaussian mixture model

ABSTRACT

Intelligent machine-tools generate a large amount of digital data. Data mining can support decision-making for operational management. The first step in a data mining approach is the selection of relevant data. Raw data must, therefore, be classified into different groups of contexts. This paper proposes an original contextual classification of data for smart machining based on unsupervised machine learning by Gaussian mixture model. The optimal number of classes is determined by the silhouette method based on the Bayesian information criterion. This method is validated on real data from four different machine-tools in the aerospace industry. Manual data mining and k-fold cross-validation confirm that the proposed method provides good contextual classification results. Then, several key performance indicators are calculated using this contextual classification. They show the relevancy of the approach.

KEYWORDS:

• Industry 4.0
• machining
• unsupervised machine learning
• GMM
• BIC

Acknowledgments

The authors acknowledge the financial support of the French National Research Agency on the ANR SmartEmma project (ANR-16-CE10-0005). They would also like to thank the contributions made by the industrial partners.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was supported by the Agence Nationale de la Recherche [ANR-16-CE10-0005].