Technology data bases for systematic evaluation of production data
Automated systems for the collection and analysis of machine, tool and quality data contribute to the enhancement of product and process quality. Frequent reference is made to the “Single Source of Truth” in the context of Industry 4.0. All relevant production data are filed once, in structured form – completely free of any redundancy. Only when this has been achieved, is it possible to conduct detailed and purposeful data analyses.
The Fraunhofer IPT develops and implements systems of this nature for a range of technologies and manufacturing methods. Interactions and dependencies within the whole manufacturing chain are revealed using appropriate data analysis software and potentials for optimization are derived as illustrated by the example of a technology database for the manufacture of replicative optics.
The technology database for the precision molding of optics contains information relating to all processes up and down stream such as the preparation of the forming tools via machining processes, tool coatings, quality analyses of the optic and of the forming tool decoating. This is achieved by recording all relevant product and process parameters along with their quality indicators in the technology database. The information is connected and filed clearly in the form of relational data structures – fully in accordance with the principle of a “Single Source of Truth”. A user-friendly front-end permits historical data records to be swiftly retrieved via filter functions. In order to identify patterns and dependencies within the process chain, a standardized SQL database with data-mining software such as “Rapid Miner” is used to evaluate these data records.
Thus optimum parameters, process conditions and process strategies for increasing the efficiency of manufacturing and product quality can ultimately be derived from neural networks, decision trees or correlation analyses and fed back into the system. The technology database and the subsequent analysis operation permit end-to-end data acquisition, holistic analysis of production data throughout the process chain and the derivation of optimum process settings. In comparison with the outcomes of conventional approaches such as Design of Experiments (DoE), the basis and quality of the data available for the identification and analysis of process dependencies and optimum parameters are considerably more wide ranging and detailed.