DBIS

Background

Power systems are so-called systems of systems. Physical assets interact through physical means as well as through information networks, as a whole most literally powering society. The study of power systems relies heavily on simulations, as the physical subjects of study are rarely accessible to research. The nature of power systems as system of systems makes the approach of co-simulation conducive to effective modeling. The various subsystems of a power system are modeled in separate simulators, which may also be used stand-alone, and are placed in a co-simulation framework to comprise holistic simulations of power systems.

These simulations are used for example to test grid control procedures, by connecting a grid control system with a power grid simulation. The co-simulation approach allows for independent modeling of different components, such as the communication infrastructure, the measurement/metering equipment and the power flow simulation. Another use case are multi-energy systems which exhibit sector coupling, i.e. the transfer and storage of energy in different forms (electricity, gas, heat). Here, different sectors are modeled independently and brought together in the co-simulation.

Simulators are typically pieces of procedural software, accompanied by documentation in natural language and some interfaces for configuration (via json, command line arguments, etc.). The individual simulators can be thought of as informal specifications of a conceptualization of the subdomain they model, with varying degrees of explicitness in code, documentation or configuration. In contrast, ontologies are formal and explicit specifications of (usually shared) conceptualizations, allowing for automatic consistency evaluation and reasoning according to the specified semantics. Creating ontologies from less formal or less specific conceptualizations is called ontology generation. Ontology-based data integration (OBDI) is the linking of heterogenous data from different sources through relating corresponding ontologies with another. In OBDI, local ontologies and global ontologies are distinguished, with local ontologies relating closely to individual data sources and global ontologies being independent from concrete data sources and often more abstract.

In the scope of the national research data infrastructure for energy (NFDI4Energy), a need for an ontology for co-simulation scenarios of energy systems has been stated. [1] In [2], a methodology for ontology-based co-simulation setup is presented. An overview and categorization of existing ontology-based data integration approaches can be found in [3]. Proficient systems exist for ontology matching and merging, as mentioned in [4] in the context of ontology integration. More in-dept surveys into ontology matching are found in [5,6]. Approaches to generate ontologies from less explicit specifications have been explored [7,8].

Problem Statement
The lack of specific and formal specifications makes configuration of co-simulation scenarios often challenging. The consistency between simulators with heterogenous models of their respective subdomains cannot be checked effectively and integration with external systems (recall the use case above, providing simulated measurements to a control system) is laborious and error-prone. Configuration is mostly stated in various configuration file formats (json, yaml, csv, etc.) or through procedural code (python in the case of the prevalent co-simulation framework mosaik). The explicit specification of configuration semantics through ontologies might address the problems of consistency assessment and data integration, as well as provide opportunities in automatic configuration generation through reasoning.

Objective
This thesis will contribute to a semantic and declarative framework for the configuration of co-simulation scenarios by fulfilling two core objectives: generating ontological representations of simulator configuration and aligning those to another and to global ontologies.

Ontology generation shall take in the various forms of data that specify the conceptualization of the simulator’s subdomain and make the semantics of the simulator’s configuration explicit. The generated ontologies should allow for automated reasoning (formal validity) and adequately describe the valid configuration space of the simulator (sematic adequacy): A-box instances of the ontology shall correspond to configurations that lead to a running simulation.

Ontology alignment shall enable consistent co-simulation scenarios and interoperability to external systems (user interfaces, data bases, complex IT systems such as control systems). To this end, adequate global ontologies have to be selected according to criteria such as domain coverage, community adoption and formal expressivity. The alignment needs to provide sufficient coverage between the concepts of local and global ontologies and comprise a formally valid ontology in conjunction.

Tasks

The thesis needs to be informed by a comprehensive overview of current approaches to power system simulation, as well as ontology extraction and alignment. Based on realistic use cases from the domain, the student will compile a set of measurable requirements and quality criteria against the components of the co-simulation configuration framework to be developed in this thesis.

Based on the use cases, the student selects co-simulation components for which an automatic formalization of the implicitly specified conceptualizations are useful and viable. Starting points for the selection of components are the symmetric power flow simulator pandapower and the timeseries database InfluxDB. The former centers around an unnormalized relational datamodel which can be used to construct a more formal specification of the concepts modelled by the simulator, while the conceptualization of the concept _timeseries_ by the InfluxDB database may for example be generated from the object-oriented data model of the python client.

Existing approaches for the generation of ontologies from the source data models may be extended through the generation of constraints for semantically richer local ontologies. These can be derived from the representations of the data models themselves or other semi-explicit specifications regarding the component that are available, such as the source code or documentation.

Alignment of the local ontologies to global ontologies, allowing for consistent configuration of entire co-simulation scenarios and interoperability with external systems, shall be implemented, building on the state of the art and selecting relevant public ontologies according to criteria such as domain coverage, community adoption and formal expressivity. Examples which may be suitable for the two components mentioned prior are CGMES and SemTS.

All contributions to the co-simulation configuration framework shall be meaningfully evaluated. Evaluation methodology should be informed by the state of the art as well as the practical application at hand. The evaluation of the generation process of the local ontologies could for example be performed with regard to the fit of the produced ontologies to the valid configuration space of the co-simulation components.
Simulations would be run with configuration corresponding to A-box instances of the ontologies, minimizing the share of simulation runs that run into errors. Another evaluation method might be the stability of the process under version changes of the component, i.e. can ontologies for different simulator versions be produced with the same quality with the same automatic pipeline. An approach for the evaluation of ontology alignments is examining the coverage of concepts by the mapping in both directions and the validity under reasoning for the merged ontology.

References
1. J. S. Schwarz et al., “Towards an Ontology for Co-Simulation Scenarios of Energy Systems.” Zenodo, Mar-2025.
2. M. Torlini, P. R. Mazzarino, D. S. Schiera, L. Barbierato, L. Bottaccioli, and E. Patti, “An Ontology-Based Methodology to Assist Complex Energy Co-Simulation Setup,” in 2025 IEEE International Conference on Environment and Electrical Engineering and 2025 IEEE Industrial and Commercial Power Systems Europe (EEEIC / I&CPS Europe), 2025, pp. 1–6.
3. F. J. Ekaputra, M. Sabou, E. Serral, E. Kiesling, and S. Biffl, “Ontology-based data integration in multi-disciplinary engineering environments: A review,” Open J. Inf. Syst., vol. 4, no. 1, pp. 1–26, 2017.
4. I. Osman, S. Ben Yahia, and G. Diallo, “Ontology integration: Approaches and challenging issues,” Inf. Fusion, vol. 71, pp. 38–63, Jul. 2021.
5. J. Portisch, M. Hladik, and H. Paulheim, “Background knowledge in ontology matching: A survey,” Semant. Web, pp. 1–55, Sep. 2022.
6. E. Thiéblin, O. Haemmerlé, N. Hernandez, and C. Trojahn, “Survey on Complex Ontology Matching,” Semant. Web, vol. 11, no. 4, pp. 689–727, Aug. 2020.
7. D. Van Assche, T. Delva, G. Haesendonck, P. Heyvaert, B. De Meester, and A. Dimou, “Declarative RDF graph generation from heterogeneous (semi-)structured data: A systematic literature review,” Web Semant., vol. 75, no. 100753, p. 100753, Jan. 2023.
8. P. Armary, C. B. El-Vaigh, O. Labbani Narsis, and C. Nicolle, “Ontology learning towards expressiveness: A survey,” Comput. Sci. Rev., vol. 56, no. 100693, p. 100693, May 2025.

Prerequisites
Background knowledge of semantic web technologies, ontology and automatic reasoning
Interest and basic understanding of power systems
Programming skills in Python and optionally another programming language
Familiarity with scientific writing and research methods (literature review, requirements definition, evaluation design)