Integration of Machine Learning Methods to Calculate the Remaining Useful Life of Mandrels

Abstract Predicting the service life of tools offers many advantages in manufacturing, such as reduced maintenance and fewer machine failures. The advances in machine learning methods, representing a sub-area of AI, are constantly creating new potentials in this field. However, process data with a sufficiently high information density are required to use machine learning for service life predictions. In many instances, a primary requirement is to map the entire use of the considered tool. Therefore, this article examines the suitability of an existing process data set of a ring rolling process that maps the whole use of the tool. The considered tools are mandrels, which are used to produce rings in the hot-forming industry. In the first step, the fracture pattern of a mandrel is microstructurally investigated. Subsequently, a machine learning model is developed to examine the suitability of this dataset to predict service life in an initial feasibility study.