Scholarlink

This chapter investigates Federated Learning (FL) as a data sovereignty preserving analytics option for manufacturing supply chains. Motivated by recent disruptions and the need for inter-company insight, it addresses four questions. At first which supply-chain areas gain the most from FL, second what best-practice implementations exist, third how FL can be integrated with current infrastructures, and fourth what product-, system- and chain-level requirements must be met. A systematic review of 21 peer-reviewed sources reveals five high-value application clusters—predictive maintenance, quality control, production optimization, collaborative planning and energy-management—where FL achieves near-centralized accuracy while safeguarding data sovereignty. The study distils prerequisites encompassing sensor granularity, edge computing, data standardization, governance and legal compliance, and discusses knowledge-distillation variants that further reduce trust barriers. Overall, FL unlocks cross-factory learning without exposing proprietary data, but its success depends on robust OT/IT integration, shared semantics and incentive-aligned consortia. The chapter closes with a research agenda and a process model to guide practitioners in deploying FL for resilient, competitive and sustainable manufacturing ecosystems.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Requirements for Using Federated Learning in Manufacturing Supply Chains

  • 章节

  • 发表日期：

    2026年

查看全文
  • Jonas Kallisch,
  • Jorge Marx Gómez

摘要

This chapter investigates Federated Learning (FL) as a data sovereignty preserving analytics option for manufacturing supply chains. Motivated by recent disruptions and the need for inter-company insight, it addresses four questions. At first which supply-chain areas gain the most from FL, second what best-practice implementations exist, third how FL can be integrated with current infrastructures, and fourth what product-, system- and chain-level requirements must be met. A systematic review of 21 peer-reviewed sources reveals five high-value application clusters—predictive maintenance, quality control, production optimization, collaborative planning and energy-management—where FL achieves near-centralized accuracy while safeguarding data sovereignty. The study distils prerequisites encompassing sensor granularity, edge computing, data standardization, governance and legal compliance, and discusses knowledge-distillation variants that further reduce trust barriers. Overall, FL unlocks cross-factory learning without exposing proprietary data, but its success depends on robust OT/IT integration, shared semantics and incentive-aligned consortia. The chapter closes with a research agenda and a process model to guide practitioners in deploying FL for resilient, competitive and sustainable manufacturing ecosystems.