Integration of machine data into product lifecycle management
Runtime | 01.04.2018 - 31.03.2020 |
Funding | DFG |
Project manager and leader | Dipl.-Ing. Dipl.-Kfm.(FH) Stephan G. Arndt, MBA (former staff member) |
Project staff |
Maximilian Zschornack (former staff member) |
Dr.-Ing. Matthias Klaus (former staff member) | |
Tom Unger (former staff member) |
Project Content
In the product development phase, the increasing use of external engineering services and the creation of domain-specific models lead to heterogeneous data sets distributed across different management levels and systems. In the operation phase of machines, different data sources, including sensors and embedded systems, cause mostly distributed and also heterogeneous data stocks. Products that are becoming more and more complex and intelligent, as well as their networking, are leading to a rapid increase in the amount of data.
There is no joint management of product and machine data in the development and operating phases. In particular, no barrier-free and consistent data exchange between machine manufacturers and operators is possible. Machine manufacturers lack access to information on real machines. Machine operators cannot access the product model and the associated development results and current changes. This circumstance is particularly problematic in view of networked machines. On the one hand, the properties and behaviour of such cyber-physical systems can only be modelled inadequately. The modelling success is limited by the dynamic and event-dependent communication between different machines and other system components that are still unknown during product development. On the other hand, primarily organisational deficits prevent a coordinated, multidirectional and reproducible data exchange.
Product lifecycle management strategies need to be extended by the proper integration of machine data: The linking of product and machine data must be realised through an integration approach that fulfils the basic requirements for the application of static methods and modern algorithms for data analysis. The quality of the analysis results as precise statements about interactions between machine components and their parameters increases. The proposed project creates a database that can be used jointly by development engineers, service technicians, data analysts, machines and other actors or objects from the development and operating phases of machines.
The project focuses on the consideration and fundamental development of the field model. The field model is a superordinate management and control unit and provides the basis for a mutual exchange of data. A field structure anchored in the field model serves the model-based and data-driven merging of product and machine parameters. The mapping of the field structure to concrete machines requires the support of field methods and field processes, which are also the subject of the field model and are being developed in the project. Field methods serve to select the correct product and machine parameters for the application. Field processes realise the derivation of the field structure based on the product model. The field model and its elements provide a qualified and consistent database for the correct application of statistical methods and modern algorithms. Tools such as machine learning are made usable for mechanical engineering.
The necessary product and machine data for the database associated with the project objective originate from product data management, maintenance, repair and operations as well as live data from the machine's own and networked information objects such as sensors and actuators. Specially designed algorithms keep the database up to date and consistent by