Machine Learning Simplifies Industrial Laser Processes A Laser With "Eyes" and a "Brain"

Related Vendors

EMPA Eidgenössische Material- Prüfungs-und Forschungsanstalt

FAULHABER Drive Systems

Wuxi InfiMotion Technology Co., Ltd.

Metal processing using lasers enables the automated and precise production of complex components, whether for the automotive industry or for medicine. However, the common methods require elaborate preliminary testing. Researchers at Empa in Thun are using machine learning to make laser processes more precise, cost-effective, and efficient.

Laser-based processes for metal processing are considered particularly versatile in the industry. Using lasers, components can be precisely welded or more complex parts can be manufactured using 3D printing—quickly, accurately, and automatable. For this reason, laser-based processes are used in numerous industries, such as the automotive and aerospace industries, where the highest precision is required, or in medical technology, for example, in the production of customized titanium implants.

Despite their efficiency, laser processes are technically demanding. The complex interactions between laser and material make the process sensitive to the smallest deviations—whether in material properties or laser parameter settings. Even slight fluctuations can lead to errors in production.

"To ensure laser-based processes can be used flexibly and achieve consistent results, we are working on a better understanding, monitoring, and control of laser-based processes," says Elia Iseli, research group leader in the Empa department "Advanced Materials Processing" in Thun. In line with these principles, Giulio Masinelli and Chang Rajani, two researchers from the group, aim to make laser-based production methods more cost-effective, efficient, and accessible—using machine learning.

Evaporate or Melt?

The two researchers first focused on additive manufacturing, specifically the 3D printing of metals using lasers. This process, technically known as "Powder Bed Fusion" (PBF), works somewhat differently from conventional 3D printing. Thin layers of metal powder are melted by the laser at precisely the right spots, so the finished component is gradually "welded out."

With PBF, complex geometries are possible that can hardly be realized using other methods. However, before production can begin, an elaborate series of preliminary tests is almost always required. This is because laser processing of metal, including PBF, fundamentally operates in two modes: In the so-called "Conduction Mode" or heat conduction welding, the metal is merely melted. In the "Keyhole Mode" or deep welding, it is partially vaporized. The slower "Conduction Mode" is suitable for thin and very precise components. Deep welding is slightly less precise but much faster and also suitable for thicker workpieces.

Where exactly the boundary between these two modes lies depends on a variety of parameters. For the best quality of the finished product, the correct settings are needed—and these vary significantly depending on the material being processed. "Even a new batch of the same base powder can require completely different settings," says Masinelli.

Fewer Trials for Better Quality

Normally, a series of tests must be conducted before each batch to determine the optimal settings for parameters such as scan speed and laser power for the respective component. This consumes a lot of material and requires supervision by a specialist. "Many companies cannot afford PBF at all for this reason," says Masinelli.

Masinelli and Rajani have optimized exactly this series of trials using machine learning. To do this, they used data from optical sensors that are already integrated into the laser machines. They taught their algorithm to "see" during a trial, based on this optical data, which welding mode the laser is currently operating in. Based on this, the algorithm determines the settings for the next trial. This reduces the number of required preliminary trials by around two-thirds—while maintaining the same quality of the final product.

"We hope that with our algorithm, even non-experts will be able to use PBF devices," concludes Masinelli. For the algorithm to be used in the industry, it would only need to be integrated into the firmware of laser welding machines by the equipment manufacturers.

Real-Time Optimization

PBF is not the only laser process that can be optimized using machine learning. In another project, Rajani and Masinelli focused on laser welding—but went a step further. They not only optimized the preliminary trials but also the welding process itself. This is because, even with optimal settings, laser welding can be unpredictable, for example, when tiny defects on the metal surface come under the laser beam.

"Influencing the welding process in real-time is currently not possible," says Chang Rajani. "That exceeds the capabilities of human experts." The speed at which data must be analyzed and decisions made is even a challenge for computers. For this reason, Rajani and Masinelli used a special type of computer chip for this task, a so-called Field-Programmable Gate Array (FPGA). "With FPGAs, we know exactly when they will execute a command and how long the execution will take—which is not the case with a conventional PC," explains Masinelli.

Nevertheless, the FPGA in their system is also connected to a PC, which serves as a kind of "backup brain." While the specialized chip is busy observing and controlling the laser parameters, the algorithm on the PC learns from this data. "When we're satisfied with the algorithm's performance in the virtual environment on the PC, we can transfer it to the FPGA and instantly make the chip smarter," explains Masinelli.

The two Empa researchers are convinced: Machine learning and artificial intelligence can contribute much more in the field of laser processing of metals. That’s why they continue to develop their algorithms and models and expand their range of applications—in collaboration with partners from research and industry.

Future-Proof Authentication with Universal RFID Readers

Universal RFID Readers: One Solution for All Credentials