This paper presents a case study at Mercedes-Benz Vans, proving the concept of a semantic-enhanced Digital Twin to solve data integration issues in product development across tools and teams. But in contrast to metadata-level interoperability methods, this approach enables a value-level solution. A multi layered knowledge graph represents product structure, behavior, and geometry, making hidden values (e.g. wheel position within Unity scene) accessible and synchronizable across systems. The exploratory case study shall prove the feasibility of the semantic-enhanced Digital Twin. The concept was tested in a suspension system use case synchronizing two Unity scenes. Therefor an artifact based on a novel framework defining the underlying processes, mechanisms and system architecture was developed. The implemented system was then tested on several exemplary use cases for basic functionality. The study confirms that the semantic-enhanced Digital Twin approach is technically feasible. Value-level synchronization of structural and behavioral data was achieved, demonstrating the potential to replace traditional file-based export/import processes. The approach supports seamless integration across engineering systems, enabling simultaneous development workflows. Overall, the novel concept and the developed prototype shall provide initial insights into the use of Knowledge Graphs for the synchronization of deep product data in digital engineering. It was the aim to lay the groundwork for future research towards a holistic engineering environment where systems are not only fully interoperable, but also interconnected to enable dynamic feedback loops, such as automatically updating simulation outputs in response to geometry changes in another system.

: Die KEA-Mod-Plattform ermöglicht es, Modellierungsaufgaben mit verschiedenen Modellierungssprachen wie z.B. UML, Petri-Netzen, EPK oder BPMN durch Dozierende zu erstellen und von Studierenden bearbeiten zu lassen. Die Plattform kam in einer großen Lehrveranstaltung mit ca. 250 Studierenden zum Piloteinsatz. Die Studierenden konnten mit Hilfe der Plattform und des integrierten Modellierungswerkzeugs eine Aufgabenreihe mit Modellierungsaufgaben zu Petri-Netzen bearbeiten und einreichen. Anschließend erhielten die Studierenden automatisiert generiertes Feedback. Das Poster beschreibt die Evaluation dieses Piloteinsatzes aus der Perspektive der Studierenden und bietet erste Ergebnisse in Bezug auf die Plattform-Usability und zur wahrgenommenen Lernförderlichkeit des Feedbacks.

Digital twins enable better control and provide a better understanding of automated manufacturing environments. However, some parts of production processes are not fully covered, as they are still carried out manually. In this paper, we present an approach for monitoring manual parts of an assembly and commissioning process efficiently. We combine scenery classification with an inventory control system based on high-precision scales. We use a head-mounted mixed-reality display to acquire images of the process context and to provide assistance to the worker. Hence, heterogeneous sensor information is combined to capture human activities and object interactions to create an accurate digital representation of actual process behavior. Additionally, sensor information is selected dynamically based on the context. The system was implemented and tested in a laboratory environment. For production environments, this provides a starting point to close a significant gap regarding digital twins in semi-automated manufacturing processes and enables context-aware worker assistance.

Recently, Artificial Neural Networks (ANN) have shown high potential in the area of Natural Language Processing (NLP). In the area of sentence compression, the application of ANNs has proven to outperform existing rule-based approaches. Nevertheless, these approaches require a decent amount of training data to achieve high accuracy. In this work, we aim at employing ANNs to derive process model labels from process descriptions. Since the amount of publicly available pairs of text and process model is scarce, we employ a transfer learning approach. While training the compression model on a large corpus consisting of sentence-compression pairs, we transfer the model to the problem of deriving label descriptions. We implement our approach and conduct an experimental evaluation using pairs of process descriptions and models. We found that our transfer learning model keeps high recall while losing performance on precision and compression rate.

The rise of Industry 4.0 and the convergence with BPM provide new potential for the automatic gathering of process-related sensor information. In manufacturing, information about human behavior in manual assembly tasks is rare when no interaction with machines is involved. We suggest technologies to automatically detect material picking and placement in the assembly workflow to gather accurate data about human behavior. For material picking, we use background subtraction; for placement detection image classification with neural networks is applied. The detected fine-grained worker activities are then correlated to a BPMN model of the assembly workflow, enabling the measurement of production time (time per state) and quality (frequency of error) on the shop floor as an entry point for conformance checking and process optimization. The approach has been evaluated in a quantitative case study recording the assembly process 30 times in a laboratory within 4 h. Under these conditions, the classification of assembly states with a neural network provides a test accuracy of 99.25% on 38 possible assembly states. Material picking based on background subtraction has been evaluated in an informal user study with 6 participants performing 16 picks, each providing an accuracy of 99.48%. The suggested method is promising to easily detect fine-grained steps in manufacturing augmenting and checking the assembly workflow.

In the competitive context of agile innovation cycles, it is necessary for companies to construct their business model leading them to a creative strategy innovation. There already are a number of methods to create business models, many of which are also implemented in software. However, these are often unstructured, unguided and static, resulting in diverse and heterogeneous business models. This complicates automated evaluation allowing recommendations. The aim is to develop a question-based tool yielding for comparable and, thus, analyzable business models based on a developed standardized taxonomy. The questions guiding through the configurator were derived from this taxonomy. A tool was developed implementing the questionbased concept. User tests were conducted as part of an evaluation showing promising results concerning usability in addition to the already achieved standardization.

The automated identification and analysis of human activities in a manufacturing context represent an interesting challenge to support the workers, providing novel solutions for managing and optimizing existing manufacturing processes. If on one hand real-time coupling of events and activities is a relatively easy task for activities which are executed by means of information systems, on the other hand, the coupling of events and human physical activities remains an unsolved problem. In this paper we present a novel paradigm based on the integration of a light-weight, low-cost body sensor network and a software solution based on machine learning for tracking working operations. This enables the fast identification of inconvenient ergonomic behaviors and process aspects, which are objective of workflow analysis, process improvement and optimization activities. To assess the usability and functionality of the system a study under real conditions was conducted in the logistic plant of a big automobile manufacturer.

Digital transformation has long since become an important topic not only for large companies but also for SMEs. However, it is becoming increasingly difficult to keep track of new technologies, assess their benefits for the company and its business model. Frequently, companies hope for cost reductions and time savings. So why not learn from projects that have already successfully introduced and applied new technologies and come into contact with the companies that carried out that project. The idea is to offer electronic consulting services by means of an information platform at low cost, flexibility and transparency. For this purpose a platform to support companies in their digitalization process is being developed that offers them multiple entry points for business transformation and / or technology innovation, to explore the potentials of new concepts and to find partners to transform their business.

The analysis of manufacturing processes through process mining requires meaningful log data. Regarding worker activities, this data is either sparse or costly to gather. The primary objective of this paper is the implementation and evaluation of a system that detects, monitors and logs such worker activities and generates meaningful event logs. The system is light-weight regarding its setup and convenient for instrumenting assembly workstations in job shop manufacturing for temporary observations. In a study, twelve participants assembled two different product variants in a laboratory setting. The sensor events were compared to video annotations. The optical detection of grasping material by RGB cameras delivered a Median F-score of 0.83. The RGB+D depth camera delivered only a Median F-score of 0.56 due to occlusion. The implemented activity detection proofs the concept of process elicitation and prepares process mining. In future studies we will optimize the sensor setting and focus on anomaly detection.

In a continuously changing business environment and the era of digitalization, business models need to adapt constantly to allow organizations to differentiate themselves from their competitors and to secure their economic survival. However, organizations are neither able to review their business model management nor systemize it productively. Hence, the combination of Industry 4.0, business model and business model management aspects emphasizes an organization’s potential and results in an increased competitive and operational success. To guide an organization’s advancement, a maturity model for business model management is developed, which delivers assistance suitable to an organization’s requirements and strategic orientation. It assesses the organization’s current maturity level and proposes sequential steps to advance towards a refined business model and process mastery by indicating improvement potentials. Thus, the maturity model links an organization’s existing organizational and operational knowledge to new concepts and makes it accessible through a modified business model for Industry 4.0.

Unternehmen passen ihre Geschäftsmodelle an, um im Markt bestehen zu können. Grund dafür ist häufig die Digitalisierung von Produkten und Dienstleistungen. Unternehmen können durch die digitale Transformation erheblich profitieren, gleichzeitig werden sie aber auch vor große Herausforderungen gestellt. Geschäftsmodelle helfen ihnen dabei, die Geschäftsideen abzubilden sowie Innovationen zu planen. Es existieren viele Methoden, um Geschäftsmodelle abzubilden, zu optimieren und zu evaluieren, wobei einige dieser Methoden auch in Software-Tools umgesetzt sind. In diesem Beitrag werden bestehende kostenfreie Software- Werkzeuge zur Modellierung von Geschäftsmodellen vorgestellt und deren Nutzen untersucht. Dadurch sollen Anwender befähigt werden, das passende Tool für ihren Geschäftsmodell-Transformationsprozess auszuwählen.

Multivariate time series classification has been broadly applied in diverse domains over the past few decades. However, before applying the classification algorithms, the vast majority of current studies extract hand-engineered features that are assumed to detect local patterns in the time series. Therefore, the efficiency and precision of these classification approaches are heavily dependent on the quality of variables defined by domain experts. Recent improvements in the deep learning domain offer opportunities to avoid such an intensive hand-crafted feature engineering which is particularly important for managing the processes based on time-series data obtained from various sensor networks. In our paper, we propose a framework to extract the features in an unsupervised (or self-supervised) manner using deep learning, particularly stacked LSTM Autoencoder Networks. The compressed representation of the time-series data obtained from LSTM Autoencoders are then provided to Deep Feedforward Neural Networks for classification. We apply the proposed framework on sensor time series data from the process industry to detect the quality of the semi-finished products and accordingly predict the next production process step. To validate the efficiency of the proposed approach, we used real-world data from the steel industry.