Digital Twins Using Semantic Modeling and AI

The digital twin approach has been increasingly applied in industrial production for several years [1], enabling the modeling, monitoring, simulation, and prediction of critical system states. AI techniques, particularly machine learning, are used to automatically capture system behavior.

However, AI-derived insights in Industry 4.0 often remain isolated, addressing only specific aspects of a production process. The identification of overarching patterns across the entire production process and product life cycle is typically hindered by the absence of a comprehensive model for the semantic and ontological classification of these insights.

From a statistical perspective, conventional AI and machine learning (ML) methods primarily generate probabilities for the occurrence of specific patterns or correlations, without identifying causal relationships. This inherent uncertainty poses a significant challenge, particularly in complex and critical applications such as industrial production. Approaches from the field of eXplainable AI (XAI), which focus on the automatic description and explanation of ML models [2], can help address this issue. For example, XAI techniques can clarify why a model predicts machine failure and which input parameters contribute to this prediction.

Even explanations and interpretations of AI-derived insights benefit from semantically rich models. Therefore, AI findings can only be fully leveraged in the context of Industrie 4.0 when integrated into comprehensive models that encompass function, behavior, structure, and geometry. This integration enables both the explanation and the transparent prediction of system behavior.

The aim of this project is to develop AI-powered, self-learning, and self-explanatory digital twins that autonomously adjust to real system behavior, ensuring an optimal representation of the production process. Insights derived from AI methods are incorporated into a semantically rich overall model, facilitating the interpretability and explainability of AI models, as well as enabling complex analyses and forecasts through simulation techniques.

Digital twins and AI in production

Digital twins were first introduced by [1] in 2003. Initially, this technology was envisioned as a means for monitoring a product or production plant. However, today, digital twins represent dynamic, virtual entities that serve as digital counterparts of physical systems, enabling complex simulations, analyses, and predictions of the system’s states.

For further processing and analyzing of collected performance and execution data, techniques from machine learning are increasingly being employed, including artificial neural networks, deep neural networks, hidden Markov models, and approaches from physics-informed machine learning. These methods are used to automatically identify typical execution patterns and correlations, as well as to enhance the prediction of future system states [3-7].

Recent research and development efforts in modeling physical systems and representing execution and performance data are increasingly focused on the use of semantic web technologies, such as ontologies and knowledge graphs, to create expressive product and production models [8, 9].

Artificial intelligence applications often face the challenge of uncertainty, particularly when the underlying model only partially represents the observed section of reality. This issue is especially critical in applications such as production, where the explainability of AI models is essential for their reliability and acceptance [2]. The term eXplainable AI (XAI) was introduced by DARPA in 2017 to describe methods for automatically explaining AI models. These approaches either aim to explain the model as a whole or replace it with a more comprehensible surrogate model [10]. XAI techniques are increasingly being applied in the Industry 4.0 domain [11, 12].

Architecture of AI-based digital twins

The core element of the AI-based digital twin is a novel, self-learning engineering model. It begins with a product and production model that represents function, behavior, and product geometry. Graph-based design languages, such as the Unified Modeling Language [13, 14] and the Design Cockpit 43 [15] (Figure 1, top), are employed. This approach integrates semantic modeling into the product and production models, enabling the ontological classification of terms and entities within the broader domain context and mapping their relationships and interdependencies.

In contrast to traditional product and production models, the semantically rich model not only acts as an expressive input for artificial intelligence and machine learning processes but also serves as a repository for the knowledge gained through AI regarding cause-and-effect relationships and influencing factors (Figure 1, right). The semantically rich model, as a vital component of the architecture, is technically implemented as a knowledge graph, modeled using the Web Ontology Language (OWL) and stored in the graph-based database GraphDB (Figure 1, middle). This knowledge graph functions as the central knowledge base for all information relevant to or generated for AI-based digital twins.

This framework is well-suited for the integrated, life cycle spanning representation of cross-domain product and process data across areas such as function, behavior, structure, and geometry. By utilizing sensors and actuators (for example, integrated through a production planning and control system), performance and execution data from the production process are collected and semantically annotated, linking them to the product and production model. AI techniques from the field of machine learning are then applied to analyze this data using the semantic model, generating insights into causal relationships and process patterns in the form of AI models.

These insights are then integrated into the overall model as part of a self-learning digital twin and serve as input for simulating production processes. Utilizing techniques from eXplainable AI (XAI), an automatic explanation and description of the AI models is generated, such as characterizing influential features or applying layer wise relevance propagation. The simulation of system behavior, along with the explanation of the underlying AI models, forms the interface of the digital twins for the end user, such as a production manager or employee.

A key concept of the proposed approach is the decomposition of the overall system into Functional Mockup Units (FMUs) [14, 16], each of which has its own dynamics and behavior. The overall system behavior is then derived from the interaction of all FMUs, which are linked through output/input relationships.

This decomposition is already modeled in Design Cockpit 43. A central aspect of the present architecture is the dynamic modeling of the behavior of individual FMUs, not solely through fixed rules or mathematical equations (i.e., classical behavior models), but by automatically learning it from execution and performance data using machine learning (ML) methods. For the FMUs defined in Design Cockpit 43, corresponding ML models can be trained and stored in the knowledge graph for further use.

All relevant aspects of the product and production models, i.e., the FMU definitions and the product and production geometries, as well as the ML models and explanations in addition to the classic simulation and behavioral models, are transferred to the co-simulation based on INTO-CPS (Integrated Tool Chain for Model-based Design of Cyber-Physical Systems) [16]. When simulating system behavior, it is possible to use different ML models to calculate the FMU output or to provide explanations for the output calculated based on the ML model. Figure 1 shows the overall architecture of the digital twin with all components and their interaction.

Validation using a FESTO laboratory system

In this study, a FESTO laboratory system is utilized as a specific application for the exemplary implementation and validation. Figure 2 (top) illustrates the various modules of the FESTO laboratory system. Following the co-simulation concept outlined earlier, the system modules are first decomposed into individual Functional Mockup Units (FMUs), and their output-input interactions are defined (Figure 2).

Validation is demonstrated using the Functional Mockup Unit (FMU) component slide in the “testing” module as a case study. To validate the concept of the ML-based, self-learning digital twin, relevant input parameters for the successful sliding process were defined for the FMU component slide, including factors such as air pressure in the slide, component weight, and the curvature of the component’s underside. The output of the FMU is the time required for the sliding process and the success of the process itself.

To validate the approach, all components of the architecture for the component slide application were implemented prototypically. Execution data was generated through test runs, system behavior was learned from this data, and ultimately, the system behavior was simulated in the co-simulation framework as part of the digital twin.

Figure 3 (left) shows a surrogate model (i.e. a substitute model for explaining arbitrary black box models such as artificial neural networks) in the form of a decision tree, which represents the learned system behavior of the component slide based on the execution data created.

Figure 3 (right) illustrates a section of the graphical simulation of the digital twin with the FMU component slide. The decision tree serves as the basis for simulating the system’s behavior, which can thus be predicted and simulated for different input parameters.

The FESTO laboratory system replicates an entire production process, comprising numerous production steps and processing operations. This makes it particularly well-suited for validating the co-simulation concept, where the overall process is decomposed into sub-units known as Functional Mockup Units (FMUs). Additionally, each module of the system is managed and monitored by a programmable logic controller (PLC). The generated execution and performance data are therefore representative of real-world systems.

The proposed co-simulation approach enables not only the detailed modeling of individual FMUs—either through mathematical/physical models or AI-based approaches—but also the representation of complex concurrent processes. However, it should be noted that, in the scope of this validation, FMUs were only linked sequentially. The modeling of complex, concurrent interactions between FMUs using AI was not within the focus of this study.

A novel approach to AI-based, self-learning digital twins

This AI-driven approach replaces traditional physical or mathematical simulations of Functional Mockup Units (FMUs) with machine learning models, eliminating the need for extensive parameter determination. By incorporating techniques from eXplainable AI (XAI), the methodology enhances the interpretability and explainability of ML models. The introduction of a semantically rich model as a core architectural component extends conventional product and production data models, enabling the structured representation of knowledge and relationships.

This model serves both as an input for AI-driven analysis and as a repository for the insights generated through ML and XAI methods. The proposed approach was prototypically implemented and validated using a FESTO laboratory system. Selected system stations were modeled, analyzed, and simulated as Functional Mockup Units (FMUs), enabling the verification of both individual components and the overall architecture. This validation demonstrated the feasibility of AI-based self-learning digital twins and confirmed the functionality of the co-simulation framework.

The next steps in the ongoing research include the integration of additional eXplainable AI (XAI) techniques, with a particular focus on natural language descriptions of machine learning models using large language models (LLMs) and retrieval-augmented generation (RAG) [17]. Furthermore, the overall approach will be tested on a real production plant to assess its scalability and practical applicability. Another promising direction involves leveraging synthetic simulation data generated from process models in Design Cockpit 43 to further enhance model training and validation.

This article was written as part of the project “KIDZ – AI-based digital twin”, funded by the Carl Zeiss Foundation in the funding program “AI breakthroughs”.

You might also be interested in

Optimized Manual Processes in Automotive Production A module-based approach for the efficient creation of work system simulations Barbara Brockmann, Tobias Jurk, Beate Stoffels, Jochen Deuse

In the manufacturing industry, the integration of digital human models into the product development and manufacturing process is becoming increasingly important. Particularly in assembly, which is characterized by a high proportion of manual tasks, motion simulations enable a realistic representation of human work and thus make a significant contribution to the evaluation of motion economy, process validation, and efficiency improvement. However, widespread application in production planning faces various challenges, such as the high initial effort required to create human simulations as well as volatile planning conditions. This article presents a practice-oriented solution from the automotive assembly sector that enables the creation of simulations with reduced effort as well as their early and consistent use in the planning process.

Industry 4.0 Science | Volume 42 | 2026 | Edition 3 | Pages 48-55

Digital Twin Technology and Architecture A synthesis of concept and practice Arka Mukherjee , Shibaji Chandra

Digital twins are a key enabling technology of the fourth industrial revolution, integrating physical systems with their digital counterparts to create intelligent, data-driven environments. This conceptual/practice-oriented paper examines how to establish a modern architectural framework for digital twins leverages modern tech-stack like IoT, Data Fabric, AI/ML, seamless integration and enterprise grade security. The paper is grounded in an abundance of literature by leading vendors and analysts in space. It offers a comparative study of different vendors implementing the solution stack in the proposed architecture.

Industry 4.0 Science | Volume 42 | 2026 | Edition 3 | Pages 114-122

Developing Virtual Reality in Learning Contexts Navigating efficiency, content relevance and scalability Stella Kanatouri , Oliver Sosna , Alexander Kulik, Sina C. Truckenbrodt , Friederike Klan , Christian Erfurth

While virtual reality can facilitate hands-on learning, its development faces barriers, including high costs and time demands and scalability challenges. This article presents two case studies that illustrate strategies for overcoming such barriers when training the next generation of skilled workers in environmental technologies. By examining approaches for streamlining development and increasing content relevance and scalability, we highlight lessons learned for future practice. We conclude by envisioning a future in which educational institutions can flexibly and cost-effectively prototype virtual reality in learning contexts, ensuring alignment with curricular goals and learners’ needs.

Industry 4.0 Science | Volume 42 | Edition 3 | Pages 26-34 | DOI 10.30844/I4SE.26.3.3

Immersive Human Digital Twins for Industry 4.0 Supporting adaptive human-centric production by integrating cognitive and physical states Tajbeed A. Chowdhury , Martina Lehser , Eric Wagner , Paul Motzki

The rapid advancement of immersive technologies has created new opportunities to transform human-machine collaboration in industry. This paper presents an immersive platform with a digital twin that combines both physical and cognitive characteristics of human dynamics. By integrating multimodal sensing, human biomechanics, and cognitive state into digital twin technology, the proposed system enhances operational safety and ensures better ergonomics. The main argument is that human digital twins are not only desirable but essential for next-generation industrial systems. We discuss the limitations of existing human modeling approaches, outline the conceptual foundations of human digital twins, and demonstrate their industrial relevance across safety, productivity, ergonomics and sustainability.

Industry 4.0 Science | Volume 42 | 2026 | Edition 3 | Pages 6-13 | DOI 10.30844/I4SE.26.3.1

Digital Twins for Emission Reduction Ex-ante case study on a pump test bench in industrial production Felix Bischoff, Ingela Tietze , Peter Hertweck, Nina van Hasz

Digital twins are frequently referred to as a promising approach for reducing greenhouse gas (GHG) emissions in industrial production; however, robust empirical evidence of their benefits under real-world conditions is largely lacking. In this case study, the emission reduction potential of a digital twin—as a conceptually described target system—is quantified ex-ante via the example of a test bench for hydraulic pumps. To this end, the GHG emissions of the original test plan for the year 2025 are determined based on actual measured energy consumption of the tested pumps and time-resolved grid electricity emission intensities. This is followed by a rule-based rescheduling, in which energy-intensive test processes are shifted to time intervals with lower emissions. The rescheduling takes operational constraints into account so that processes and equipment remain unchanged. The savings potential is determined by comparing the GHG emissions of the reference and the optimized case.

Industry 4.0 Science | Volume 42 | 2026 | Edition 3 | Pages 16-24 | DOI 10.30844/I4SE.26.3.2

Experiencing Digital Twins in Production and Logistics The fischertechnik® Learning Factory 4.0 as a development platform for possible expansion stages Jan Schickram, Tareq Albeesh, Deike Gliem , Sigrid Wenzel

The fischertechnik® Learning Factory 4.0 has proven to be a suitable experimental environment for testing digital twins. Depending on the targeted maturity stage, the functions of a digital twin range from status monitoring and forecasting to the operational control of production and logistics systems. To systematically classify these functions, this article presents a maturity model that serves as a framework for the development of a digital twin. Building on this, selected use cases are implemented in a test and development environment based on a system architecture with multi-layered logic structure. These initial implementations serve to highlight application purposes, relevant methods, and typical challenges and potentials in the transfer to real factory environments.

Industry 4.0 Science | Volume 42 | Edition 2 | Pages 30-37 | DOI 10.30844/I4SE.26.2.30