High-tech industry moving steadily towards end-to-end digitalization with Knowledge Graph
(Article first published in October 2023 on d1g1talAgenda)
Headquartered in Singapore and with a large number of offices globally , ASMPT is the world’s leading provider of hardware and software solutions for semiconductor and electronics manufacturing. ASMPT offers Semiconductor Assembly and Packaging as well as SMT (Surface Mount Technology) Solutions: from wafer coating to various solutions for assembly and packaging of sensitive electronic components in a wide range of consumer devices, including electronics, mobile communication, computer, automotive, industrial and LED (displays). ASMPT’s close collaboration with its clients and the company’s continuous investment in research and development contribute significantly to ASMPT offering innovative and cost-effective solutions and systems that enable users to achieve higher productivity, higher reliability, and improved quality.
The end-to-end data communication, automation, and high-speed processing of chip dies are currently the biggest challenges for the semiconductor industry. As a global technology leader with more than 12 000 employees for electronics production, ASMPT responds with end-to-end solutions under the motto ‘Enabling the Digital World’.
Bringing together what belongs together
The basis for ASMPT’s end-to-end digitalization are latest-generation IT backbone systems. Since the migration of data from old systems to the new PLM system is a very complex undertaking, ASMPT went in search of a system provider for competent support. With the Linksphere platform from CONWEAVER GmbH from Darmstadt, Germany, ASMPT has found a technology that makes it possible to combine and harmonize data from a heterogeneous system infrastructure across domains with tried-and-tested efficiency on an industrial scale and thus make it ‘migration-ready’.
As part of a proof of concepts, the focus was on the data of the current product generations of ASMPT’s Semiconductor Solutions segment. Old data from legacy systems was filtered out and not included in the further analysis.
The solution included the networking of data sources, automated detection of data discrepancies, and different bill-of-material (BOM) structures for the same issue in the different data sources. In addition, there was generated a data layer for the future upload to the PLM system and options for further machine identification of discrepancies as well as further machine networking and data harmonization.
For this purpose, huge amounts of database entries were loaded into Linksphere’s so-called Digital Brain — here we are talking about several million data sets! The data sets were then made available to the ASMPT colleagues for further processing.
Conclusion
A total of more than 2.1 million instances, 71 million relations, and 130 million attributes from four vaults (SAP, Teamcenter and two legacy systems) were loaded into CONWEAVER’s Digital Brain and checked. Numerous discrepancies between SAP and the legacy systems were uncovered and their correction for the subsequent migration to the new PLM system was taken into account. The data provided made it possible to significantly reduce the differences in BOMs. The solution was implemented while the company was still running. ASMPT’s day-to-day business continued unaffected.
