Deep Learning For Real-Time Quality Control
The International Center for Networked, Adaptive Production (ICNAP) comprising of three Fraunhofer Institutes will be presenting an AI-supported analysis system for visual quality control at the SPS trade fair in Nuremberg, Germany from November 12 to 14, 2024. ICNAP will demonstrate how deep learning can be used to check weld seams on battery cell mod-ules for defects in real time.
The use case that the researchers will be demonstrating shows the automated detection of defects in weld seams on battery cell modules. To do this, the battery cell module is manually placed under a camera and a photo is taken, which is then analyzed by a pre-trained deep learning model. During ongoing production, the artificial intelligence independently detects defects what would otherwise have to be identified by trained specialists. The new system, on the other hand, immediately displays the defect detection in color in real time – red for a defective product and green for a flawless product. This allows defects to be detected during the production process and rejects to be avoided at an early stage.
The system specifically inspects battery cell modules, but it can also be applied to numerous other products that require a visual surface inspection. The process increases production efficiency through consistent and objective quality assurance and relieves skilled workers of this routine task. By automating the inspection process, companies can also counter future staff shortages.
The system’s error rate continues to fall over time thanks to a growing volume of training data. Data analysis can take place either directly on site on the shop floor or using external cloud systems. During the SPS, the researchers use their new ‘FCTRY CLD’, a local real-time capable cloud from the Fraunhofer IPT. This enables the operation of time-critical applications for production, such as the control of machines or the processing of sensor data. The advantage: the data is processed securely in the production environment and does not leave the company, as is the case with public cloud providers.
ICNAP: A strong network for the digital future
The International Center for Networked Adaptive Production, ICNAP, is a collaboration between the three Aachen-based Fraunhofer Institutes for Production Technology IPT, for Laser Technology ILT and for Molecular Biology and Applied Ecology IME and 24 renowned companies worldwide. ICNAP was founded as an open research community that develops and tests new approaches to digitalization in production as an industrial test environment.
For more information: www.ipt.fraunhofer.de