Given the 350 production changeovers per day, a portfolio containing roughly 1,200 different products, and 17 million Simatic components produced per year, about 50 million items of process and product data need to be evaluated and used for optimization in order for production at the Siemens Electronics Works Amberg (EWA) to run smoothly. In addition, groundbreaking technologies like artificial intelligence (AI), Industrial Edge computing, and a cloud solution are already enabling highly flexible and extremely efficient and reliable production sequences.
Industrial Edge computing and AI for increased throughput
“With Edge Computing, data can be immediately processed where it’s generated, right at the plant or machine,” says Dr. Jochen Bönig, Head of Strategic Digitalization at Siemens Amberg. This is what EWA is doing, for example, on the production line where PCBs are manufactured for components of the distributed I/O.But even here, production isn’t sufficiently optimized, and it’s neither the fault of plant availability nor process quality. The bottleneck is at the end of PCB production, at the automatic x-ray inspection section.Circuit boards of the size of a fingernail accommodate function-related BUS connectors with various connecting pins. In a non-integrated test, the soldered joints of these connecting pins are x-rayed and checked for correct functioning. Should another x-ray machine be purchased for about €500,000? (Click here to read an expert article on the subject at the Siemens Blog.) The alternative is artificial intelligence. The data from the sensors is transferred to a cloud via the TIA (Totally Integrated Automation) environment, which consists of a controller and an Edge device. Experts train an algorithm that’s based on AI and the process parameters. The algorithm learns how process data reflecting the quality of the soldered joints behaves and controls a model that runs on an Edge application at the plant.“The model predicts whether or not the soldered joints on the PCB are free of faults: in other words, whether or not an end-of-line test is necessary. Thanks to closed-loop analytics, this data can immediately be factored into production,” explains Bönig .



