Polymer manfacturer REHAU deploys Fujitsu AI software to automate high quality checks

Polymer producer REHAU Industries, which makes plastic seals and componentry for doorways and home windows, is about to roll-out an industrial AI inspection software from Fujitsu on its manufacturing traces in Germany following profitable subject trials. The brand new AI answer, based mostly on unspecified “good applied sciences”, replaces visible high quality inspections, and roots-out 99 % of product defects, the corporate has stated. The implication is it makes use of machine-vision digicam expertise, and works on non-public 5G networks, as required. The pilot is to be transferred right into a full manufacturing system, stated REHAU; additional use instances will probably be added as properly. 

REHAU Industries, a part of REHAU Group, is predicated in Rehau in Higher Franconia. It sells artificial polymer merchandise for the automotive, building, furnishings, supplies, medical and industrial sectors, usually used for componentry in home windows and inside fittings, and cooling and fluid options. Fujitsu was drafted-in to work on an automatic high quality management system for manufacturing of plastic ‘extrusion profiles’ (steady shapes created by a technique of extrusion), the place high quality assurance is a problem due to the vary of merchandise and the demand for “excellent floor”.

A press release defined: “REHAU produces greater than 200,000 totally different variations of those in varied colors, patterns, and designs in a number of factories worldwide. The profiles are manufactured utilizing the extrusion course of on giant machines and wound into coils on the finish of the road… Visible inspection by line employees [has until now] introduced plenty of challenges, together with the truth that high quality of products being manufactured may solely be checked in the beginning or finish of a bundle, or by taking samples throughout manufacturing.”

It went on: “Regardless of excessive course of stability and course of management, this introduced an ongoing danger of undetected errors and high quality management points. To deal with these challenges, Fujitsu delivered a brand new high quality management system to streamline this course of in pilot operations, throughout which it efficiently detected over 99 % of all defects on a check knowledge set (99.32 %). The built-in AI was skilled utilizing photographs of the manufactured profiles in optimum situation, which permits the system to reliably detect all types of defects throughout the manufacturing course of.”

Fujitsu stated it’s utilizing machine studying to doc errors and enhance fault detection. “Potential defects and anomalies might be anticipated on the premise of the data gained and thus averted,” it stated. The pair stated the co-creation course of between the 2 corporations was efficient and the event was speedy. A totally automated visible inspection answer is to be deployed throughout manufacturing of all of REHAU’s extrusion profiles. The Japanese agency will present the answer alongside a bunch of parallel tech at Hannover Messe in Germany on the finish of April.

Niels Strohkirch, member of the administration at Fujitsu Germany, stated: “High quality administration is more and more vital to clients. As anybody skilled with manufacturing processes can let you know, even small manufacturing errors can have a significant affect down the road, particularly if you’re making an attempt to optimize using supplies. With our AI answer, we’re demonstrating how skilled employees wielding good applied sciences can obtain higher leads to areas the place human guide or visible inspection alone has been the norm.”

Tobias Lehner, good applied sciences engineer at REHAU, stated: “With the brand new high quality management answer, we can take product high quality to a brand new stage, decreasing the danger of complaints and waste. This contributes to each price discount and general, extra sustainable manufacturing. As the following step, we now plan to switch the pilot right into a productive system – and ideally discover additional use instances.”