Federated-Learning-based Decision Support for Industrial Internet of Things (IIoT)-based Printed Circuit Board Assembly Process

• Published in: Journal of grid computing Vol. 20; no. 4; p. 43
• Main Authors: Tsang, Y. P., Wu, C. H., Ip, W. H., Lee, C. K. M.
• Format: Journal Article
• Language: English
• Published: Dordrecht Springer Netherlands 01.12.2022; Springer Nature B.V
• Subjects: Assembly; Automation; Circuit boards; Communication; Computer Science; Decision support systems; Federated learning; Industrial applications; Industrial Internet of Things; Intelligence; Machine learning; Management of Computing and Information Systems; Neural networks; Printed circuit boards; Printed circuits; Printers; Process parameters; Processor Architectures; Productivity; Quality management; Solders; Subject specialists; User Interfaces and Human Computer Interaction; Printed circuit board; Federated learning; Industrial internet of things; Assembly; Decision support
• ISSN: 1570-7873; 1572-9184
• DOI: 10.1007/s10723-022-09637-8

Solder paste printing (SPP) is one of the critical processes for the printed circuit board assembly to reliably apply the solder paste on raw PCBs for the component placement through surface mount technology. Although computerised SPP machines have been developed in the past few years, the reliance on domain experts cannot be neglected to fine-tune corresponding process parameters so as to maintain productivity and quality. This study exploits federated learning on the industrial internet of things (IIoT) paradigm to establish an intelligent decision support system across various networked machines. The IIoT-based squeegee blade is deployed in the SPP machines for better machine-to-machine communication and interconnectivity. In contrast, a global machine intelligence model is aggregated in a decentralised and privacy-preserving manner. Consequently, the automated and sustainable manufacturing management for PCBA is achieved, where wastes from trial production runs are eliminated.