Drilling is one of the most commonly used machining processes. Despite the relatively simple kinematics of the process, monitoring becomes more complex due to the depth of penetration into the workpiece. Whilst milling or turning usually produces stable signal waveforms, process-specific effects during drilling can lead to signal jumps. Conventional methods such as envelope analysis usually employ fixed thresholds that are exceeded by such signal jumps. To avoid false alarms, these thresholds are often set quite broadly. However, this reduces sensitivity to critical faults. Furthermore, envelope analysis requires reference data from identical, previously machined workpieces. In single-part production, however, such data is generally not available.
These limitations are addressed by AutoBohr. The solution developed in the project consists of several coordinated modules. First, the system autonomously detects the current phase of the drilling process, such as during the air cut, during the actual drilling operation, or upon exiting the workpiece. On this basis, the respective process stage is specifically evaluated. This phase-dependent analysis allows processes to be assessed regardless of their duration and the selected drilling cycle.
In a second module, the process is evaluated using historical data to distinguish expected signal jumps from critical anomalies. Through the use of neural networks, this approach is also extended to processes for which no historical data is available. This means that even the very first drilling operation on a new workpiece can be reliably monitored, even if the selected spindle speed, feed rate or drill size have never been used before. The system evaluates the ongoing process in real time and automatically detects deviations.
The approach was tested at a contract manufacturer under real production conditions. The analysis of production data showed that anomalies such as tool breakage are reliably detected, whilst maintaining a high degree of robustness against varying process parameters. With AutoBohr, an important step has thus been taken towards flexible and intelligent process monitoring during drilling. Particularly for companies manufacturing individual parts or small batch sizes, the developed method offers new possibilities for ensuring quality and process reliability from the outset, without lengthy training phases or extensive reference datasets.
This project was funded by the Federal Ministry for Economic Affairs and Energy (BMWE) following a resolution by the German Bundestag. The IFW and iba AG would like to express their gratitude for this support.
Contact:
For further information, please contact Kirill Tkachuk on +49 511 762 18382 or by email at tkachuk@ifw.uni-hannover.de.
