Managing datasets from sheet metal forming systems using the self-describing file format HDF5

Abstract Although collecting data, more precisely resource data, on field-level from sheet metal forming systems is an important and time-consuming step towards designing new process chains, managing these datasets remains challenging. Field-level data typically consists of absolute values collected from sensors or actuators and must be enriched with contextual information on the data collection procedure, including detailed experiment descriptions and specifications of the recording sensors. In industrial practice, the collected field-level datasets, experiment descriptions, and sensor specifications are stored in separate files as plain text with comma-separated values, requiring manual assignment of field-level data and their data collection properties. Permanently linking and organizing field-level datasets and data collection information in a single file would significantly simplify the management and exchange of datasets from sheet metal forming systems. This dataset consolidation could reduce the number of separate, yet content-related, files, ensure the dataset’s integrity, and eliminate the need for manual data assignment. Self-describing file formats, like the Hierarchical Data Format version 5 (HDF5), offer a promising solution for this data management challenge, as they support the link between descriptive information and large data tables in a single file, making them self-descriptive. Currently, they are not well established in sheet metal forming, so an HDF5-based concept is contributed to organize field-level datasets, experiment descriptions, and sensor specifications, and store them together in a single file. The concept is applied to datasets from a backward extrusion forming system, a representative system in sheet metal forming.