Two companies I’ve known for more than 30 years in my editing career have joined forces to build a wide range of machines that will include an automated machine learning system—that when applied to new machines—will make it possible for end users to fine tune machines to top performance and use automated condition monitoring, allowing both the machine builder and end user to maintain and keep equipment running without any surprise breakdowns. In fact, adding this functionality to a machine opens a new business model for both the machine builder and the end user.

Weidmüller has decades of experience in creating I/O solutions for machine builders. These include industrial network connectivity solutions, terminal blocks, enclosures, cable marshalling solutions, wireless connectivity, PLC interfaces, signal conditioning equipment, measuring and monitoring systems, remote I/O, tools, and the list now includes a new software product, Automated Machine Learning Software.

This new software product caught GEA’s attention, and the machine builder intends to expand and enhance its range of services in terms of systems and plants. GEA has already initiated a pilot at its Oelede site.

Industry 4.0 a challenge and great opportunity

Digital technologies (sometimes known as Industry 4.0, IIoT and other terms) provide challenges for machine builders and companies in the mechanical and plant engineering sector, but these technologies also provide completely new opportunities—both for machine builders like GEA and end users as well. To meet today’s production needs, plants and their equipment must be quickly adaptable to individual products and customer requirements.

As we’ve seen in FE articles, preventive maintenance is giving way to predictive and prescriptive maintenance, which means getting intelligent and actionable data from machines’ condition monitoring systems to keep them running at peak efficiency. “We have been dealing with the topic of condition monitoring and condition monitoring of machines and have set up threshold value analyses for quite some time. But we were aware that the potential of this topic is much greater,” says Kerstin Altenseuer, head of service development at GEA. “In the long term, the aim was to map processes or to be able to optimize applications together with our customers. And, of course, also to establish new business models and areas of application such as leasing or subscription models for our machines.”

GEA, with over 125 years of expertise in the manufacture of separators and decanters for the separation of liquids, benefits enormously from this experience. These machines are used in various industries such as the food industry, chemicals, pharmaceuticals, biotechnology, and more. With new business models or applications, the company intends to open up new sources of revenue. “However, we realized relatively quickly that we needed the expertise and help of data experts for this project,” says Altenseuer. “It is not easy to identify and recruit the appropriate experts, i.e. data scientists, even if a company like GEA would in principle have good ‘cards.’ But we would have needed several, which doesn’t make things any easier.”

Looking outside one’s own four walls

GEA realized the solution it needed was external to the company. In its search for a solution to this problem, GEA became aware of Weidmüller and the company’s expertise in the field of industrial analytics through the “It’s OWL” excellence cluster. The goal: to think differently about the service offering for GEA customers and establish a smart services program. At the same time, GEA realized it could improve the quality and performance of its machines, and it needed to develop new business models to position GEA competitively in the market.

GEA and Weidmüller first worked on how the project could be set up and what the central goal would be. “It quickly became clear that we would first prove the feasibility in a proof of concept and then enable GEA to develop and operate (machine learning) ML models independently,” says Tobias Gaukstern, business unit head of industrial analytics at Weidmüller. With the help of the Automated Machine Learning Software Service, the experts at GEA should be able to train machine learning algorithms and statistical models independently.

“The AutoML tool simplifies and accelerates the application of ML for application experts, without the expert knowledge in ML being necessary,” says Gaukstern. Mechanical engineers often face the problem that their design, automation and process experts cannot easily transfer their knowledge into machine learning solutions. How do you bundle this application knowledge in software or even in an algorithm?

“The solution was very interesting for us because we have many process engineers who know the machines very well and can interpret the data to a certain extent. With Weidmüller’s help we can now transfer this knowledge into an algorithm,” explains Matthias Heinrich, manager digital solutions at GEA. A proof of concept (PoC) with historical data was carried out in Oelde to check how the theoretical considerations can be applied on site in the production environment at GEA. The aim was to automatically detect anomalies in the behavior of separators in the dairy industry.

Close cooperation wins the day

That the project was a complete success was also due to the good and close cooperation within the teams. On one hand, the regional proximity was a great advantage, because the project team was able to get together easily and quickly to discuss individual aspects. “Weidmüller also has a very broad data scientist perspective,” says Altenseuer. “At the same time, as a mechanical engineer you feel well understood because you don’t just sit down with IT specialists but with engineers who know the machines.”

Within the scope of the project, GEA provided the input and the requirements, and Weidmüller then implemented the proof of concept. “This division of labor has proven to be very effective. We had regular, good coordination and very good results, which formed the basis for the pilot application and finally the transfer served the series,” says Gaukstern.

The applications were used in connection with an existing IIoT scenario for condition monitoring at GEA. “Everyone is talking about digitization. But in the end, we want to provide added value. We want to take the next step with Weidmüller’s solution,” says Altenseuer. “Until then, a number of tasks still need to be completed, such as improving data connectivity and data quality, before we can get started. So far we have connected 500 machines in the existing portal and we want to transfer Weidmüller’s solution to these machines as quickly as possible,” explains Altenseuer. And she looks to the future: “I also see great potential for transferring the new technology to other areas at GEA.”

For more information, visit or