Institute of Automation and Information Systems, Technical University of Munich, 85748 Garching, Germany.
Sensors (Basel). 2021 Jan 22;21(3):745. doi: 10.3390/s21030745.
Data collection from distributed automated production systems is one of the main prerequisites to leverage information gain from data analysis in the context of Industrie 4.0, e.g., for the optimization of product quality. However, the realization of data collection architectures is associated with immense implementation efforts due to the heterogeneity of systems, protocols, and interfaces, as well as the multitude of involved disciplines in such projects. Therefore, this paper contributes with an approach for the model-driven generation of data collection architectures to significantly lower manual implementation efforts. Via model transformations, the corresponding source code is automatically generated from formalized models that can be created using a graphical domain-specific language. The automatically generated architecture features support for various established IIoT protocols. In a lab-scale evaluation and a unique generalized extrapolation study, the significant effort savings compared to manual programming could be quantified. In conclusion, the proposed approach can successfully mitigate the current scientific and industrial challenges to enable wide-scale access to industrial data.
从分布式自动化生产系统中进行数据采集是利用 Industrie 4.0 中数据分析信息增益的主要前提之一，例如，用于优化产品质量。然而，由于系统、协议和接口的异构性，以及此类项目中涉及的众多学科，实现数据采集架构与巨大的实施工作相关联。因此，本文提出了一种模型驱动的数据采集架构生成方法，以显著降低手动实现工作。通过模型转换，相应的源代码可以从使用图形领域特定语言创建的规范化模型中自动生成。自动生成的架构支持各种已建立的工业物联网协议。在实验室规模的评估和独特的广义外推研究中，可以量化与手动编程相比节省的大量工作。总之，所提出的方法可以成功地缓解当前的科学和工业挑战，从而实现对工业数据的广泛访问。
