What's it good for?
Intelligent machines are able to make quality forecasts of the components they are currently producing. They independently analyze sensor data and parameters from the machine control system, enabling them to supply valuable information about faults in the production process.
In tests, researchers at Fraunhofer IPA have thus obtained reliable forecasts about the thickness, impermeability and defects of a weld created by an ultrasonic welding machine making high-quality welds. Machine learning methods for forecasting quality can also be applied directly to other production processes, such as extrusion, injection molding, deep-drawing, tumble clinching or joining processes.
What is that good for?
The self-learning analysis tool "Smart System Optimization" detects deviations from the norm. It independently identifies and reports errors and their cause in running production processes, as well as throughout production and assembly lines.
What is new?
The process now not only functions reliably on the basis of image and video data is also capable of analyzing the high-frequency data streams occurring in the machine control systems involved.