Unsupervised machine learning framework for early machine failure detection in an industry

Abstract

This article describes a credible and prognostic analysis in this study for failure detection of a machine in the industry. An interpretable methodology and an informative functionality are portrayed, trained with the dataset and their explicatory implementation is compared and evaluated. In this paper, we will design a deployable end-to-end grading model to forecast whether or not the machine will fail. We will train state-of-the-art algorithms for gradient enhanced decision trees (GBDT) and compare their performance.

Subject Classification:

• 68Txx
• 68M11
• 68-XX

Keywords:

• Industry
• Predictive maintenance
• Machine learning