The Augsburg AI production network-DLR Lightweight Production Technology Center (ZLP), Fraunhofer IGCV and the University of Augsburg-use ultrasonic sensors to correlate sound with the quality of composite material processing.
An ultrasonic sensor installed on a CNC milling machine to monitor the quality of machining. Image source: All rights reserved by the University of Augsburg
The Augsburg AI (Artificial Intelligence) production network-established in January 2021 and headquartered in Augsburg, Germany-brings together the University of Augsburg, Fraunhofer, and research on casting, composite materials and processing technology (Fraunhofer IGCV) and the German lightweight production technology center. German Aerospace Center (DLR ZLP). The purpose is to jointly research artificial intelligence-based production technologies at the interface between materials, manufacturing technologies and data-based modeling. An example of an application where artificial intelligence can support the production process is the processing of fiber-reinforced composite materials.
In the newly established artificial intelligence production network, scientists are studying how artificial intelligence can optimize production processes. For example, at the end of many value chains in aerospace or mechanical engineering, CNC machine tools process the final contours of components made of fiber-reinforced polymer composites. This machining process puts high demands on the milling cutter. Researchers at the University of Augsburg believe that it is possible to optimize the machining process by using sensors that monitor CNC milling systems. They are currently using artificial intelligence to evaluate the data streams provided by these sensors.
Industrial manufacturing processes are usually very complex, and there are many factors that affect the results. For example, equipment and processing tools wear quickly, especially hard materials such as carbon fiber. Therefore, the ability to identify and predict critical wear levels is essential to provide high-quality trimmed and machined composite structures. Research on industrial CNC milling machines shows that appropriate sensor technology combined with artificial intelligence can provide such predictions and improvements.
Industrial CNC milling machine for ultrasonic sensor research. Image source: All rights reserved by the University of Augsburg
Most modern CNC milling machines have built-in basic sensors, such as recording energy consumption, feed force and torque. However, these data are not always sufficient to resolve the fine details of the milling process. To this end, the University of Augsburg has developed an ultrasonic sensor for analyzing structure sound and integrated it into an industrial CNC milling machine. These sensors detect structured sound signals in the ultrasonic range generated during milling and then propagate through the system to the sensors.
The structure sound can draw conclusions about the state of the processing process. “This is an indicator that is as meaningful to us as a bowstring is to a violin,” explained Prof. Markus Sause, director of the artificial intelligence production network. “Music professionals can immediately determine from the sound of the violin whether it is tuned and the player’s mastery of the instrument.” But how does this method apply to CNC machine tools? Machine learning is the key.
In order to optimize the CNC milling process based on the data recorded by the ultrasonic sensor, the researchers working with Sause used so-called machine learning. Certain characteristics of the acoustic signal may indicate unfavorable process control, which indicates that the quality of the milled part is poor. Therefore, this information can be used to directly adjust and improve the milling process. To do this, use the recorded data and the corresponding state (for example, good or bad processing) to train the algorithm. Then, the person operating the milling machine can react to the presented system status information, or the system can react automatically through programming.
Machine learning can not only optimize the milling process directly on the workpiece, but also plan the maintenance cycle of the production plant as economically as possible. Functional components need to work in the machine as long as possible to improve economic efficiency, but spontaneous failures caused by component damage must be avoided.
Predictive maintenance is a method in which AI uses collected sensor data to calculate when parts should be replaced. For the CNC milling machine under study, the algorithm recognizes when certain characteristics of the sound signal change. In this way, it can not only identify the degree of wear of the machining tool, but also predict the correct time to change the tool. This and other artificial intelligence processes are being incorporated into the artificial intelligence production network in Augsburg. The three main partner organizations are collaborating with other production facilities to create a manufacturing network that can be reconfigured in a modular and material-optimized manner.
Explains the old art behind the industry’s first fiber reinforcement, and has an in-depth understanding of new fiber science and future development.
Post time: Oct-08-2021