Scholarlink

<p>This study investigates the development of a data-centric process model tailored to address the specific challenges associated with the manufacturing of Micro-electro-mechanical systems (MEMS)-based sensors. These challenges include managing complex data flows, sensitivity to environmental factors, and the need for high precision in production. MEMS-based sensor production is a critical area in advanced manufacturing, yet conventional data science solution lifecycles provide only limited explicit guidance for the real-time adaptivity and tightly coupled calibration processes required in this context. The research aims to create an interdisciplinary solution for real-time anomaly detection, adaptive calibration, and context-sensitive feature selection. Using the Design Science Research (DSR) methodology, the study integrates design requirements, design principles, and design features into a nine-phase process model. The development of the artefact involves collaboration across roles, such as, domain experts, data scientists, data engineers, and project managers, to ensure practical applicability in industrial settings. Validated through three case studies: wafer map clustering, adaptive calibration, and smart feature selection, the proposed process model significantly reduces production bottlenecks, enhances accuracy, and optimizes resource allocation. Overall, the study contributes a novel, structured process model that addresses critical demands in MEMS manufacturing and illustrates how a DSR-grounded, domain-specific lifecycle can be instantiated in practice, while providing design knowledge that may inform tailored AI process models in other high-stakes, sensor-intensive production contexts.</p>

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

A process model to facilitate AI-based sensor manufacturing optimization

  • 期刊论文

  • 发表日期：

    2026年5月7日

查看全文
  • Itilekha Podder,
  • Udo Bub,
  • Robert Winter

摘要

This study investigates the development of a data-centric process model tailored to address the specific challenges associated with the manufacturing of Micro-electro-mechanical systems (MEMS)-based sensors. These challenges include managing complex data flows, sensitivity to environmental factors, and the need for high precision in production. MEMS-based sensor production is a critical area in advanced manufacturing, yet conventional data science solution lifecycles provide only limited explicit guidance for the real-time adaptivity and tightly coupled calibration processes required in this context. The research aims to create an interdisciplinary solution for real-time anomaly detection, adaptive calibration, and context-sensitive feature selection. Using the Design Science Research (DSR) methodology, the study integrates design requirements, design principles, and design features into a nine-phase process model. The development of the artefact involves collaboration across roles, such as, domain experts, data scientists, data engineers, and project managers, to ensure practical applicability in industrial settings. Validated through three case studies: wafer map clustering, adaptive calibration, and smart feature selection, the proposed process model significantly reduces production bottlenecks, enhances accuracy, and optimizes resource allocation. Overall, the study contributes a novel, structured process model that addresses critical demands in MEMS manufacturing and illustrates how a DSR-grounded, domain-specific lifecycle can be instantiated in practice, while providing design knowledge that may inform tailored AI process models in other high-stakes, sensor-intensive production contexts.