Subscribe Button 1

Self-Learning In-Process Quality Control For High-Precision Machining

The SonicShark system from Hufschmied Zerspanungssysteme GmbH is used for inline quality control and process monitoring during milling processes. In cooperation with nebumind an extension of the system was implemented. The previously purely visual quality control of a worker is expanded into an automated, self-learning monitoring. With each captured workpiece, the extended system reports rejects ‘inline’ and constantly optimizes its recognition, independently of adjustments in the milling process. The self-learning defect detection opens up new potential in manufacturing to save time and money in quality assurance.

The innovative SonicShark technology uses structure-borne noise sensors and an adaptive AI to system recognizes anomalies in production processes, identifies material inhomogeneities and ‘hears’ the onset of tool wear. Due to the inline control that accompanies the machining, SonicShark saves time and money in quality assurance and enables more efficient tool use and predictive maintenance.

“There are many solutions mapping manufacturing data onto a workpiece. But until now, none of these solutions could perform calculations with the data and automatically give recommendations for action. The software from nebumind is unique in this respect and, as an extension of the SonicShark system, will make quality assurance much more efficient in the future“ states Ralph Rudolf Managing Director Hufschmied Zerspanungssysteme GmbH.

Increasing Efficiency In Quality Assurance

For its quality control during the process, the tool manufacturer Hufschmied Zerspanungssysteme GmbH
has developed a new system inspecting quality inline. Nowadays, the actual milling process is so much optimized in terms of required time, that only little savings potentials is left here. The efficiency potential has rather shifted to quality assurance.

Searching For More Efficient Quality Assurance

The structure-borne sound sensor of the intelligent SonicShark system evaluates vibrations and acoustic signals in the machine tool room to monitor process, workpiece and tool quality. The sensor records the data time-based in combination with machine data such as position and feed values. The aim of Hufschmied Zerspanungssysteme GmbH is to make such quality assurance even more efficient: On the one hand, the defect detection of the SonicShark system shall take place inline. On the other hand, the time required to machine, evaluate and report an a not OK component (n.o.k) shall be significantly reduced. Automated acquisition and evaluation of sensor data is indispensable for this.

Locating Defects Right On The Workpiece During Machining

For an automated evaluation of the sensor data on the workpiece, the nebumind software was used. The software evaluates sensor data not only time-based, but above all location-based, and can thus locate defects precisely in the component. The worker is automatically shown where a defect has occurred in the workpiece. This way, an n.o.k. component can be identified already during the milling process and replaced right away. In addition, the time required by the operator to find and evaluate the defect is reduced to a minimum.

Improving Defect Recognition With Every New Defect

Through the evaluation of new, so far unknown, defects by the worker, a self-learning process takes place in the software. With each newly manufactured workpiece, the software becomes more intelligent – only relevant defects are reported and immediately unloaded.

Another special feature of the software is that the automatic evaluation of sensor data works independently of the process control. This means that a worker can change the milling speed for different workpieces or start the milling process in different areas of the workpieces without disturbing the self-learning quality control. This makes the manufacturing process much more flexible.

Saving Time and Resources

The recording of CO2 emissions during the process is also enabled through the SonicShark system, which can also lead to adjustments in the process. Since on average 25 – 30 % of manufacturing costs are caused by quality assurance and testing, there is very large savings potential here.

For more information: