Keywords: Data Analytics, Data Integration, Industrial Data Quality, Data Analysis, Robust Regression, Feature Selection, Predictive Asset Management, Obsolescence Management, Digital Companion, Railway Infrastructure
Astract: The increasing extension of railway networks worldwide entails highly complex settings of signaling systems. For maintenance planning tasks, decision support approaches are required as analyzing data of differently structured sources may be extremely time consuming. Although existing data analytics techniques have proven to be able to extract and analyze large amounts of data in different fields of application, in the railway domain more advanced approaches are needed to tackle an interplay of complex industrial data settings. This includes aggregating data from different formats, identifying ambiguities, and visualizing most important information. To support maintenance decisions, a modern approach has to be able to process data in a coherent and consistent manner while minimizing the time required for interventions. In this industrial thesis, we present a procedural approach that combines new techniques for processing configuration and operational data of signal systems to support decision-making in maintenance tasks. Since configuration and operation data reveal different data structures and formats, in the first step the approach ensures high quality data integration into a data warehouse. As storing duplicate or contradicting information may have business-critical effects, an interactive technique provides an efficient process where the user resolves ambiguous data classifications. Once data is stored in the data warehouse, information of hardware components and its properties, i.e., features, can be used as variables. This allows the following technique to build a regression model to estimate the quantity of components based on a set of input features, but also ensures the identification of relevant features related to a hardware component. The resulting regression model is combined with a stochastic model to predict the number of hardware components needed for existing and planned systems in the future. Instead of showing plain prediction results, we propose advanced visualizations to support technical engineers to quickly grasp all important information including the uncertainty of the prediction. Extending the mere predictions, we propose the concept of a digital companion which prescriptively recommend maintenance actions in system configurations. All our findings have been evaluated in continuous collaboration with experts of the railway domain. Ultimately, the techniques developed have been integrated into the railway business which confirms the relevance and usefulness of our work.

Keywords:
Astract: The COVID-19 pandemic did not only dramatically impact the personal and social lives, for many academics, it also demanded immediate changes to the way their courses are taught. While a pragmatic approach is to do conventional lectures via video streaming platforms, much more may be done to educate students also in a remote setting properly. This particularly holds true for practice-oriented and technology-engaging courses. This paper describes our experience of transforming an in-person Master level class on model-driven software engineering into a distance learning one. We describe the structure, the content, the teaching and examination format, and the used platforms in detail. We critically reflect on our experiences and report the feedback gained by a post-class student evaluation. We believe this paper provides meaningful lessons learned and best practices for other educators challenged with the task of teaching similar courses in a remote setting. With this paper, we publish an openly available Github repository that features all course content including sample solutions for all practical lab assignments.

Keywords: Model Engineering, MDE, Model Virtualization, Model Synchronization, Model Integration, Constraint repair, Xtext, Ecore, OCL
Astract: Multiple teams working on a single system may each have different viewpoints, and thus, use different models. These models may have partly shared, unique, or interrelated information, requiring model integration. To work faster and in a more parallel way, temporary inconsistencies between multiple models may be accepted. However, shared information only edited by a single team could still be immediately made known globally. The two main approaches to model integration are model virtualization, i.e., deriving all models from a single source of truth and model synchronization, i.e., propagating changes between different materialized models. While model virtualization does not allow temporary inconsistencies between models, model synchronization may require storing duplicate information redundantly, even if only a single team is involved. Thus, this thesis combines model virtualization with model synchronization into a hybrid approach. A new model virtualization approach helps arbitrarily adding or subtracting models from a base model. The base model can be a single model, an intersection or union of multiple models, a modification of another base model, or a model derivation. As we can store arbitrary (user) changes to the base model without affecting it, we allow temporary inconsistencies and arbitrary changes to the base model, e.g., as a result of changing the derivations source model. Incompatible changes never require user intervention, but just cause semantic constraint violations in a newly defined synchronization model, which is valid if and only if all inter-model constraints including feature derivations are fulfilled. To produce quickfix suggestions in (textual) model editors, optimal model synchronization is regarded as finding an optimal synchronization model. For this optimization, both model finders and heuristic search is employed. Model derivations can be specified using a new basic model derivation language, which includes both derivation and synchronization constraints in a single model. This allows for pure derivation by not editing the derived model as well as pure synchronization by specifying constraints just for inter-model consistency, but not for derivation. This hybrid approach is feasible and can support use cases like editing multiple models simultaneously using virtualization. Our proposed model repair does significantly reduce the number of (synchronization) constraint violations and prevent new ones due to improved autocompletion as shown in our evaluation scenarios.

Keywords:
Astract: With the emergence of Cyber-Physical Systems (CPS), several sophisticated runtime monitoring solutions havebeen proposed in order to deal with extensive execution logs. One promising development in this respect is the integration oftime series databases that support the storage of massive amounts of historical data as well as to provide fast query capabilitiesto reason about runtime properties of such CPS.In this paper, we discuss how conceptual modeling can benefit from time series databases, and vice versa. In particular,we present how metamodels and their instances, i.e., models, can be partially mapped to time series databases. Thus, thetraceability between design and simulation/runtime activities can be ensured by retrieving and accessing runtime information,i.e., time series data, in design models. On this basis, the contribution of this paper is four-fold. First, a dedicated profilefor annotating design models for time series databases is presented. Second, a mapping for integrating the metamodelingframework EMF with InfluxDB is introduced as a technology backbone enabling two distinct mapping strategies for modelinformation. Third, we demonstrate how continuous time series queries can be combined with the Object Constraint Language(OCL) for navigation through models, now enriched with derived runtime properties. Finally, we also present an initial evaluationof the different mapping strategies with respect to data storage and query performance. Our initial results show the efficiency ofapplying derived runtime properties as time series queries also for large model histories.

Keywords: Modeling Languages, MDE, DSML
Astract: Model-driven development (MDD) offers advantages in the development process by raising the level of abstraction and reducing the complexity of a specific domain. Domain-Specific Languages (DSLs) used in MDD raise the productivity for developers and improve communication with domain experts. DSLs can be divided into textual languages (TLs) and graphical languages (GLs). IDEs provide different application programming interfaces (APIs) for developing tool support for various languages. If tool support for a language is required in multiple IDEs, the implementation of the same language has to be repeated for each IDE based on their different APIs. For this purpose, the Language Server Protocol (LSP) for TLs separates the editor interface from the language logic and allows the reuse of one language server implementing the language logic across several LSP-based IDEs. Like the LSP, the Graphical Language Server Protocol (GLSP) was invented to transfer the advantages of extensibility and reusability to GLs. Increased complexity in software development leads to an increased number of errors in programs and models and, therefore, it requires debugging facilities for both TLs and GLs. Therefore, the Debug Adapter Protocol (DAP) was invented to standardize the communication between the IDE and a concrete debugger. The DAP intends that an IDE offers one generic graphical user interface for debugging functionality and that there is one debugger per language implementing the language-specific debug logic that can then be reused among IDEs. This thesis analyzes a way of combining the DAP for TLs and the GLSP for GLs to support model debugging in a web-based environment. Furthermore, it is evaluated whether the well-known debugging concepts for debugging source code can be transferred to GLs. The thesis intends to reuse existing debugging components and frameworks developed for TLs. The results of this work are evaluated in two use cases. The first use case is the running example and aims at investigating whether the DAP for TLs can be reused for GLs. The second use case intends to evaluate the reusability of the developed framework concerning further GLs and domain problems. The case study’s results indicate that the DAP enables efficient multi-editor integration of debuggers, i.e., one language-specific debugger can efficiently be integrated with a DAP-based debugging interface and that a DAP-based debugging interface can efficiently integrate with multiple language-specific debuggers. Further, the results show that the developed debugging framework meets the requirements of a modern debugger and facilitates the integration of debugging support for further GLs.