Learning while welding

The internationally active Grenzebach Group and the electrical engineering and automation company Weidmüller are collaborating in the field of data analysis and predictive maintenance. The focus of their research is on the condition and quality monitoring of friction stir welding technology. Customized analysis software from Weidmüller monitors the welding process and is designed to automatically detect and record any anomalies in the welding process in the future.

Innovative Technology

Friction stir welding is a pressure welding process. The weld is not melted; enormous pressure and a rotating tool heat the material during the welding process, making it malleable. At the same time, the tool pin rotates to stir the components to be joined. It enables the joining of light metals such as aluminum, creating secure and heat-resistant, long-lasting joints.

This process is one of the most stable welded joints and is therefore mostly used on electromobility components and in critical sectors such as aerospace. Joins that are not possible with other welding technologies are achieved with friction stir welding. Another advantage of the technology is that it does not produce smoke, sparks, or toxic gases.

Dual monitoring for higher quality

The Grenzebach Group offers systems for this welding process. To support customers with quality control, the company is expanding the technology with Weidmüller's Industrial Analytics Tool "AutoML." This complements the traditional inspection of the welded piece with a data-driven approach that increases the quality of the weld. "We want our customers to produce with the best possible quality and be informed early on of any deviations," explains Christian Herfert, Senior Vice President of Special Technologies at Grenzebach.

Industrial AutoML records and evaluates sensor data during the welding process. This includes, among other things, the mechanical forces acting on the tool. The analytics tool compiles and further processes the data. After training the tool with the help of hundreds of standard welds, the system creates models capable of automatically detecting weld deviations. "Our software compares the measured values ​​with the trained welds. If the software detects a deviation outside the defined parameters, the operator knows that something is wrong," explains Dr. Daniel Kress, Data Scientist at Weidmüller.

Using a numerical identifier, Grenzebach assigns a make to each data set. This data is forwarded to Industrial AutoML so that customers can view and track the quality of each workpiece individually. Reliable documentation of deviations increases transparency for customers.

This young technology does not yet have many specialists. The employees working there should be extensive support. Defects that are not visible to the naked eye are detected by the software. These can be explicitly taken into account when inspecting the workpieces, which increases product quality. The additional documentation provided by Industrial AutoML provides double protection for Grenzebach's customers.

A partnership for greater customer benefits

The Grenzebach Group and Weidmüller are working together to ensure that the welding machine manufacturer's customers can benefit from simplified processes. In addition to improved documentation of welding processes, Industrial AutoML also enables reliable, predictive maintenance. In a further step, machine learning paradigms can be created from the generated data, which fully automatically identify critical measurements and energy-intensive processes. These can then be further developed according to user needs toward sustainable production or, as an ML solution, support quality assurance specialists.

Grenzebach is thus optimizing its machines and upgrading them to meet the increased demands of modern systems. This ensures the high quality of its products while simultaneously facilitating easy interaction between the system operator and the system. Customers can now derive added value from the data generated during their production even without a data scientist.

Weidmüller offers numerous products and solutions related to Industrial IoT and automation. "As an 'enabler from data to value,' it is our goal to enable our customers to derive added value from data for themselves and their companies," explains Dr. Thomas Bürger, Head of the Automation Products and Solutions Division at Weidmüller. Weidmüller provides its customers with innovative concepts for data acquisition, preprocessing, analysis and visualization.