Choose your country:
Leadec global
Go to Homepage

Good Vibrations: Finding a Solution during a Hackathon

No machine runs like another, even when they are identical. So, it's not easy to uncover anomalies in vibration data. An intelligent algorithm can help. Born in this year’s ICNAP hackathon.  

In an ideal production world, ideal rules apply: Identical machines operate identically, wear synchronously and their vibrations also follow certain standards. More precisely, according to ISO 10816, which specifies which standard threshold values apply to vibration sensors. But the reality is different: “Such thresholds are a good starting point when it comes to alarms,” says Andreas Zeller of the Smart Factory Group. “But they don’t reflect real-world sensor applications and installations, especially when monitoring multiple sensors on similar machines.”  


Sensors are sensitive   

This has to do with their installation position and alignment, among other things, but also entirely different factors. For example, the state of wear of the tools used in cutting machines is partly responsible for whether the machine runs “smoothly” or not. The age and degree of use of the coolant can play a role or pick-and-place movements superimposing the individual vibration signal. This overtaxes the sensitive sensors, which really only want to react reliably to a signal.  


Individual zero line instead of standard 

So how can the right data be filtered? The magic word: machine learning. This involves IT systems themselves recognizing certain patterns on the basis of existing data and ultimately developing solutions. Applied to the vibrations of machines, this means developing, verifying and deploying an algorithm that is able to generate individual threshold values using learning data from the first week after a machine has been serviced or installed. Andreas Zeller: “The task for the ICNAP hackathon was to deliver an algorithm for dynamically setting alarm thresholds. An individual zero line, so to speak.” 

A weekend of spinning heads 

For Tim Klaas of the “TUBOT” hacker team from Berlin, it was precisely this problem that held a special appeal: “We’re all absolutely enthusiastic about machine learning and knew right away that we wanted to get to grips with it.” TUBOT is made up of “TUB,” the initial letters of Technical University Berlin and “BOT” for robotics. The five-member team – consisting of four electrical engineers and one computer scientist – tested four different methods and ultimately relied on an autoencoder to solve the task. This artificial neural network first compresses the input data and then reconstructs it. This produces a reconstruction error, which is used to train the network. The particular difficulty was that it was not obvious what an anomaly looked like, since only error-free training data was available. “We therefore built the algorithm using the training data so that error-free data would be reconstructed correctly, but in the anomaly case there would be a reconstruction error. We were also able to implement the same principle much more efficiently in terms of training time using the principal component analysis,” says Klaas, a master's student. And the effort paid off: Although it was the very first hackathon for the TUBOT students, they were able to convince the Leadec jury with their solution. “We are really impressed by the precision and technical know-how with which work was done here. Now we want to implement an even further optimized variant on our installation as well,” says Andreas Zeller.  

Digital hackathon as fertile ground

Leadec has been a member of the International Center for Networked, Adaptive Production (ICNAP) since 2020 and values the platform for the exchange of experience between community members from industry and science. The hackathon has been a regular part of the Hannover Messe program  for years and even though it was held digitally, Tim Klaas takes a lot away with him: “It's always exciting to work on concrete use cases like this one from Leadec. We apply our theoretical knowledge in practice and take away new insights – certainly also for upcoming hackathons.” 


In addition to Tim Klaas, the TUBOT team includes Marvyn Bornemann, Christian Hegeler, Hannes Lorkowski and Kyra Kerz. 


You might also be interested in:

The Smart Factory Group - Leadec's Innovation Hub

Article in Leadec Pulse - What the Sensors Can Tell Us