Praline Digital

Efficient support for operating personnel in detecting and rectifying quality defects

Quality defects in products and semi-finished products repeatedly cause rejects and downtime. Detecting and removing faulty products requires a high level of manpower, which many companies are often no longer able to afford due to staff shortages.

The use of digital operator assistance systems is a future-proof solution strategy here. The assistance systems record and analyze the fault situation and then provide the machine operators with information on the cause of the fault and suitable solutions as required based on the description of the situation from a database.

By digitally supporting personnel in the sustainable elimination of the causes of faults, assistance systems can help companies to save resources and costs and successfully meet the challenges posed by the shortage of skilled workers.

Integration into existing systems

The “imageSAM” operator assistance system can be easily and cost-effectively integrated into existing systems. In order to demonstrate the concrete feasibility of such integration and the potential of the technology solution, the research project will equip an existing production line in the “chocolate pilot plant” for the manufacture of chocolates using the cold stamping or one-shot process with webcams and lighting technology and evaluate the recorded data with the help of training data sets.

This gives interested companies the opportunity to get to know how an operator assistance system works in terms of detecting and assigning errors in the manufacturing process and providing instructions on how to rectify the causes of errors on an industry-oriented production line and to see the benefits for themselves.

It's the “why” that counts!    

The use of visual systems for quality control is nothing new. However, up to now these systems have mostly only been aimed at detecting and rejecting faulty products. They do not eliminate the causes. Errors can therefore occur again and again and lead to rejects. In this case, the experience and knowledge of the operating personnel is again required in order to deduce the cause of the error from the error patterns.

As part of the “imageSAM” research project, machine learning models for classifying images with typical faults were developed at the Fraunhofer IVV and the achievable accuracy of the assignment to the causes of faults was determined.

By linking fault images with the causes, it is possible to provide operators with precisely matching troubleshooting instructions for the detected deviation, e.g. via tablet, smartphone or PC.

Project informationen