changes What’s driving the process industry

Service in the spotlight

From purchasing to logistics, Endress+Hauser already utilizes AI in many areas to create transparency and optimize processes. One application specifically helps to monitor product quality in the field.

Text: Kirsten Wörnle
Graphics: Teresa Wagner

Artificial intelligence can analyze enormous volumes of data, recognize patterns, draw conclusions, and even learn while it is doing so. Service It’s no wonder the technology is attracting such huge interest from businesses. Endress+Hauser uses the technology to analyze, improve and automate processes. Examples of its use include altering device specifications, checking weld seams or selecting the best shipping routes. With three million units coming off the production line every year, the efficiency gains are huge; AI saves time and prevents mistakes.

AI can also help out when those instruments are operating in the field. Endress+Hauser uses a cloud-based application called Product Lens to monitor in real time how the company’s measuring devices are performing worldwide. The insights are used to learn and anticipate potential problems. There is an in-house instrument database containing 60 million entries that Product Lens checks every night, identifying new service cases and cross-checking them with historical data. If there are a lot of service callouts concerning a particular type of instrument, Product Lens will analyze technician service reports and spot which cases might hint at a broader issue.

Identification is by no means straightforward because service is usually a routine part of commissioning, calibration and maintenance. So reports containing words such as ‘repair’, ‘problem’ or ‘replacement’ may indicate manufacturing-related issues – or maybe not. “In cultures like Japan’s that emphasize politeness, maintenance is sometimes documented as repair,” says Enrico De Stasio, head of corporate quality, lean and IT. “Other cultures say maintenance when they actually mean repair.” Product Lens has already learned all of these subtleties. “Context is key,” says Thomas Fricke, head of the marketing services division. “We trained our AI on 15,000 cases, and now it identifies relevant ones with a 95 percent success rate. Our in-house experts review those in greater detail and narrow them down. The tool learns more with every case it finds.”

VALUABLE INFORMATION

There haven’t been any product recalls as yet. But the application brings many extra benefits. “We can track in the field whether tiny changes to the manufacturing process bring any undesirable side effects,” De Stasio adds. All Endress+Hauser employees have access to Product Lens: its findings are just as relevant to manufacturing as they are to development and technical support. When there is a new service case, technicians can get on the right track more quickly. “If one measuring instrument is flagging up a problem at just one customer, while thousands and thousands of the same model are working fine, we’re probably looking at a faulty connection or operator error,” Fricke explains.

Published 23.08.2024, last updated 16.09.2024.