ETH researchers use artificial intelligence to improve quality management in digitized production processes. In an experiment at the semiconductor manufacturer Hitachi Energy, the proportion of defective products was halved.
Whether for complex machines, pharmaceutical products or microchips: the more expensive a product, the more crucial its quality in order to survive on the market. This is particularly true for industrial production in high-wage countries such as Switzerland. It is all the more surprising that poor quality still accounts for an average of 15 percent of operating costs in industrial production. The reason for this: an often outdated quality management system.
Outdated quality management
The manufacture of complex, industrial products requires a large number of interconnected work steps. For example, to produce semiconductors that are used in computers, trains or wind turbines, between 200 and 400 production steps are necessary. Various things can go wrong with any of these steps. For example, if the temperature in one machine is a little too high or the pressure in another is a little too low, errors can affect a large part of the products. “In semiconductor production, it is not uncommon for the error rate to fluctuate between 5 and 70 percent, as the manufacturing process is very complicated. A lot of money is at stake, ”explains ETH professor Torbjørn Netland from the Department of Production Management.
Up until now, it was very difficult to precisely identify sources of error in more complex production processes of this type. Traditional methods, which are used in numerous factories around the world for quality assurance, only ever allowed engineers to examine a few parameters at a time. In the meantime, however, thousands of related parameters are measured in complex production processes. How these interact and affect the entire production process, however, could not be analyzed using traditional methods. ETH professor Netland and his co-authors intend to change this: “We want to bring quality management into the digital age. Because the methods that many producers continue to use are often up to 100 years old, ”says Netland.
Algorithm identifies sources of error
The ETH researchers first developed an algorithm that imitates the individual steps in the production of semiconductors. They then fed the algorithm with as much historical production data as possible, such as temperature or pressure, which are measured in machines. “On the basis of this data, the algorithm learns under which conditions the quality of the semiconductors is good and when there are high error rates,” says Julian Senoner, the first author and research assistant at the Netland professorship.
The advantage of the AI-based method is that you can analyze any number of factors and relationships in the production process and uncover more complex relationships between parameters. In this way, sources of error can be identified more systematically and across the entire production process. Well-trained engineers in the factories are by no means superfluous. On the contrary: “Our algorithm primarily reveals sources of error that have not yet been discovered. How exactly these are remedied, however, still requires a great deal of technical expertise and human creativity, ”explains Netland.
Source: com! professional by www.com-magazin.de.
*The article has been translated based on the content of com! professional by www.com-magazin.de. If there is any problem regarding the content, copyright, please leave a report below the article. We will try to process as quickly as possible to protect the rights of the author. Thank you very much!
*We just want readers to access information more quickly and easily with other multilingual content, instead of information only available in a certain language.
*We always respect the copyright of the content of the author and always include the original link of the source article.If the author disagrees, just leave the report below the article, the article will be edited or deleted at the request of the author. Thanks very much! Best regards!