Machine learning for detection of anomalies in press-hardening: Selection of efficient methods

Abstract

The paper addresses machine learning methods, utilizing data from industrial control systems, that are suitable for detecting anomalies in the press-hardening process of automotive components. The paper is based on a survey of methods for anomaly detection in various applications. Suitable methods for the press-hardening process are implemented and evaluated. The result shows that it is possible to implement machine learning for anomaly detection by non-machine learning experts utilizing readily available programming libraries/APIs. The three evaluated methods for anomaly detection in the press-hardening process all perform well, with the autoencoder neural network scoring highest in the evaluation.