Collision control of machine tools and components

5G acoustic emission sensor for monitoring tool wear and tool breakage. The sensor is attached directly to the workpiece and continuously records a signal (see waterfall diagram) and sends a frequency spectrum to an AI model that classifies the data with regard to tool wear.

© Fraunhofer IPT

Mastering manufacturing processes such as the milling of highly complex components means, above all, being able to quickly detect deviations in the movement of the tool and react within a few milliseconds. This can be achieved with machine-integrated sensor technology in conjunction with 5G, as this mobile communications standard is able to transmit the acquired data quickly and reliably. This enables the machine tool to react to changes in the process before a component is damaged. The 3rd Generation Partnership Project (3GPP), a global cooperation for the standardization of mobile communications technologies, is developing a new standard for future 5G products for this very responsive connection - the so-called 'Ultra Reliable and Low Latency Communication' (URLLC).

Niels König is head of the Production Metrology department at the Fraunhofer Institute for Production Technology IPT in Aachen.

© Fraunhofer IPT

In order to be able to react even faster and more reliably to changes in the production process in the future, the Fraunhofer IPT and the mobile communications equipment supplier Ericsson are testing the URLLC functions in the collision detection use case. Ericsson is supplying the Fraunhofer IPT with a modular 5G test system that meets the required URLLC functions in this use case. The test system operates in the millimeter wavelength range (mmWave), a 5G spectrum with frequencies between 24 and 28 GHz, which can be applied for from the German Federal Network Agency with immediate effect. In contrast to stationary 5G systems in the 3.7 to 3.8 GHz frequency range, 5G systems in the millimeter wavelength range are new in production and have hardly been tested to date. However, the larger spectral range allows higher data rates of more than 10 Gbit/s and latencies <1 ms, meaning that time-critical applications in particular can be implemented reliably.

Praveen Mohanram is a research associate in the Digital Infrastructures department at the Fraunhofer IPT in Aachen.

© Fraunhofer IPT

With collision detection, a structure-borne sound sensor, which is attached directly to the workpiece or integrated into the machine spindle, detects a collision of the tool within a few milliseconds so that the machine can be stopped in time before major damage to the component or the machine can occur. In the event of a collision between the machine and component, strong acoustic emission shocks are generated, which the sensor detects. The frequency range of the acoustic emission shocks differs from conventional tool breakage detection: the frequency range and the deflections are higher, as the tool actually collides with the component.

The intelligent structure-borne sound sensor detects acoustic emissions in the range from
1 to 900 kHz. It is connected to a sensor platform that has integrated signal conditioning and can record the sound emissions and calculate the frequency spectrum. The platform runs at a sampling frequency of 1 MHz and generates a data packet with frequency spectra every few milliseconds, which is sent to an AI model integrated into the cloud for data analysis. This in turn classifies the data with regard to tool wear.

The system uses 5G communication to handle the resulting high data rates of over 12 Mbit/s and to ensure reliable communication with low latency for timely intervention and an accurate assessment of tool wear. Thanks to high data rates of up to 10 Gibt/s and latencies <1 ms, the new mobile communication standard is suitable for making manufacturing processes more mobile, flexible and adaptive than before, improving the quality of components, reducing costs and thus increasing productivity. With the 5G test system, the detected acoustic emission shocks are sent to an AI model integrated into the cloud in order to stop the process in time within a few milliseconds. Process control using an acoustic emission sensor is just one of many ways of monitoring the machine process. The existing infrastructure does not need to be changed; it is flexible and scalable.