Today's communication networks are often characterized by high technical complexity for users and operators. Heterogeneous network architectures, edge computing and network slicing as well as constantly increasing demands on quality of service, agility and flexibility cause unprecedented operational challenges. Insufficiently known control variables, parameter fluctuations and degradation effects require high safety margins in network planning and cause increased energy consumption.
Configuration errors can threaten not only network resilience, but also the information security of operational and user data. Malfunctions are often difficult to localize and can only be remedied with considerable time expenditure.
AI-NET-ANIARA aims to develop innovative solutions to these problems based on concrete application scenarios from the field of sensors and manufacturing. Intelligent, cross-site end-to-end automation at the network and service level is intended to avoid manual process steps as far as possible and enable fully autonomous network operation in the future.
Start 01.02.2021 End 31.01.2024
Funding source: BMBF https://www.forschung-it-sicherheit-kommunikationssysteme.de/projekte/ai-net-aniara
Project management: VDI/VDE Innovation + Technik GmbH
The goal of the Institute for Machine Tools and Production Technology (IWF) at the TU Braunschweig in the AI-NET ANIARA project is to identify and develop potentials for flexible production systems enabled by 5G and edge computing. As part of the project, IWF will equip the Incremental Manufacturing Lab experimental laboratory with intelligent edge devices that can process sensor and machine data close to the data origin using AI (artificial intelligence) algorithms and machine learning (ML). The results of the algorithms will be used, on the one hand, for simplified monitoring of the process by human operators (locally using dashboards, remotely using virtual reality, for example) and, on the other hand, also for direct control of the equipment. It is also being analyzed how the generated and pre-trained algorithms can be transferred to other plants at other locations. Together with Opel, it is also being investigated how 5G and edge computing technologies can be profitably applied to use cases in the context of energy efficiency and indoor air quality. Methods for the orchestration of edge devices will be developed and the transferability to real-world use cases as well as the training of AI models will be tested.
In the Incremental Manufacturing Lab (IML), automation will be advanced at both sub-process and process chain levels. To illustrate, the integration of a single edge device (ED) for the additive manufacturing sub-process of the IML is shown in the figure below. First, defined quality requirements, e.g., layer cohesion for additive manufacturing and surface finish for machining, are loaded into the edge server(ES). Based on this information, the ES can assign previously trained AI models to the individual edge devices (ED), giving them their intelligence. During operation, the individual EDs are constantly supplied with machine and sensor data, which enables them to determine the quality of a sub-process during production on the basis of the AI models and to create a so-called quality gate. This principle is transferable to all process steps, for which a large number of EDs are used.
The calculated results from the individual EDs converge again in the ES and can be used there for monitoring the entire process chain and for deriving optimized operating strategies. In the ES, the data is already highly aggregated and can thus be transferred to higher-level systems such as Manufacturing Execution Systems (MES) and Enterprise Resource Planning (ERP). In order to link several production sites (ERP level) with each other, the Fraunhofer IPT is investigating how the results can be transferred to other sites ("Global Cloud"). On the store floor level, functions for control, data backup, visualization and remote monitoring can be derived from the ES and implemented.
