Marker-free identification of components for bearing rings
Final Report Abstract
In the transfer project “Marker-free identification of components for bearing rings”, the application partner Schaeffler promotes its activities in the field of Auto-ID. By using a component-inherent marker, expensive markers such as laser marking can be saved, surface alterations and impairments can be avoided and a hidden identification marker invisible to the human eye can be employed. By using this technology, Schaeffler will be enabled to identify individual bearing rings without additional marking process and to gather, record and analyze component-related data in consecutive manufacturing steps. In addition, marker-free component identification will complement the Industry 4.0 activities of the application partner. To this end, a method for individual component identification previously developed at the Institut für Informationsverarbeitung (TNT) at Leibniz Universität Hannover was adopted to the production of bearing rings. Using this patented method, a camera image of a surface portion of a machined component is assigned a “fingerprint”, which is stored in a database along with the corresponding ID. The functionality of this “fingerprinting” method was successfully evaluated by simulation, experiments, and in the form of a demonstrator. The purpose of the transfer project is the further development and adoption of existing algorithms for the fullyautomatic identification of individual bearing rings of different diameters, ranging from 10 mm to 240 mm. Here, the ground front face acts as information-carrier area. For the intended application within the production process, several modifications were implemented. First, a complete scan of the entire fingerprinting area in every process step is too time-consuming. Instead, the entire scan needs to be performed only once at the start of the process chain. At the consecutive read-out and identification stations, only a small circular ring sector needs to be imaged to perform the matching. Second, a scalable search and matching algorithm was developed, which is capable of operating in real-time even on a database with high numbers of pieces of several 10,000 or 100,000 items. The improved image processing algorithms were implemented in a prototype installation in a productive grinding line. Finally, the system was evaluated with respect to robustness and mass-production readiness. All requirements of the industrial partner were met.
