The Fraunhofer IVV Dresden will present two new systems for efficient process management at interpack 2017 from 4-10 May 2017 in Düsseldorf, Germany.
The Mobile Cleaning Device (MCD) combines the benefits of traditional automated cleaning systems and the versatility of manual cleaning.
The MCD has an optical sensor system for dirt detection and adaptive cleaning.
Inline sensor systems can be used to provide information on the areas that needed cleaning, the completing of cleaning tasks and outcomes.
The virtual twin of the MCD comprises an adaptive model of the cleaning process.
Combining this with cognitive control concepts and the sensor system for dirt detection allows for the first time adaptive cleaning, namely cleaning adapted to the hygienic state of the machinery.
On flexibility, the movement between machine modules can take place via its drive unit or by utilizing existing transport systems such as conveyor belts.
In contrast to standard cleaning systems, the MCD is not installed in a dedicated way in a machine, rather it can be used in a versatile way to clean several machines.
Separately driven nozzles are available for foam and spray cleaning.
Besides the cleaning of whole machines, the targeted cleaning of parts of a machine is also possible.
Meanwhile, the Fraunhofer IVV Dresden will present the first concepts of self-learning assistance systems for processing machinery at the show.
The work recognizes that even the most advanced production lines are prone to often short faults/stoppages every five minutes on average.
Processes and machinery are becoming ever more complex.
Many production line operators are thus unable to remedy faults at their source, and so only manage to alleviate the short-term effects.
Even highly advanced sensor systems are not always able to prevent faults, for example those caused by fluctuating product properties.
The most important source of knowledge for remedying machine faults is the knowledge of experienced, qualified machine operators.
In order for this knowledge to be passed down to less experienced personnel, the Fraunhofer IVV Dresden is pursuing various concepts for providing operators with information about fault elimination relevant for the prevailing conditions.
These concepts have now been brought together in SAM, the self-learning assistance system for machines, which helps operators remedy faults via a quasi navigation system.
A foundation for this is anomaly detection in patterns of sensor signal based on the techniques of data mining and machine learning.
Future steps will develop a cooperative dialog system.
This will allow the assistance system to learn directly from the operator and together propose a problem-solving strategy, without the SAM itself actively engaging in the production process.
