Entwicklung eines Konzepts zur Kontextualisierung von Prozessdaten

von Trotha, Christian; Epple, Ulrich (Thesis advisor); Diedrich, Christian (Thesis advisor); Kleinert, Tobias Theodor (Thesis advisor)

D├╝sseldorf : VDI Verlag GmbH (2021)
Book, Dissertation / PhD Thesis

In: VDI Fortschritt-Berichte : Fortschritt-Berichte VDI. Reihe 20, Rechnerunterst├╝tzte Verfahren 476
Page(s)/Article-Nr.: XII, 177 Seiten : Illustrationen, Diagramme

Dissertation, RWTH Aachen University, 2021


Data is being generated and stored during the operation of process plants. This is, for example, data from planning processes (e.g. the P&ID flow chart) or process data (e.g. measurement series). Standardized interfaces, information models and exchange formats, such as the asset administration shell, OPC UA, DEXPI or PandIX are becoming increasingly popular for the exchange and access to this data. This enables clear identification and semantic referencing of the stored data. The next step to accomplish data-based decisions is the preparation and contextualization of this data. This contextualization and preparation must be carried out with little overhead and using standardized tools so that the existing data can be used easily and efficiently. This work addresses the challenge of contextualization. It describes a concept that enables consistent modeling of the data context in the process industry. The core of the work is a model hierarchy, which can be used to model the relationships between individual data points. In this way, the available information can be interconnected to form knowledge. The concept takes advantage of the wide range of available information about structural relationships, for example in the form of circuit, construction and installation plans, flow diagrams or control architectures. These structures determine the local dependencies between individual data points. Thereby the contextualization system enables the efficient search and filtering of the available data, making it easier to use them. The technological basis of the work is a knowledge representation using a semantic network. Semantic networks represent information using nodes and edges. They are particularly suitable for describing relationships and dependencies between data points and form the ideal basis for contextualization. A major challenge of using semantic networks is the lack of formal specification of terms and relationships. For the use case of contextualization of process data this challenge is addressed in this work by employing the model hierarchy mentioned above. The applicability of the concept is demonstrated by means of a prototype implementation for the lab system \pumping station". It is based on the Grakn graph database platform and uses various asset administration shells, BaSys40-components and PandIX as information sources.