MBSE-AI Integration 2025 Abstracts


Area 1 - MBSE-AI Integration

Full Papers
Paper Nr: 7
Title:

Ontology-Driven LLM Assistance for Task-Oriented Systems Engineering

Authors:

Jean-Marie Gauthier, Eric Jenn and Ramon Conejo

Abstract: This paper presents an LLM-based assistant integrated within an experimental modelling platform to support Systems Engineering tasks. Leveraging an ontology-driven approach, the assistant guides engineers through Systems Engineering tasks using an iterative prompting technique that builds task-specific context from prior steps. Our approach combines prompt engineering, few-shot learning, Chain of Thought reasoning, and Retrieval-Augmented Generation to generate accurate and relevant outputs without fine-tuning. A dual-chatbot system aids in task completion. The evaluation of the assistant’s effectiveness in the development of a robotic system demonstrates its potential to enhance Systems Engineering process efficiency and support decision-making.
Download

Paper Nr: 12
Title:

Transforming Systems Engineering in Nuclear Projects with Generative AI: A Path to Efficiency and Compliance

Authors:

Jérémy Bourdon, Julien Rodriguez, Quentin Lesigne, Pauline Suchet, Berenger Fister, Loic Montagne, Olivier Malhomme, Lies Benmiloud-Bechet and Robert Plana

Abstract: This article explores the integration of generative artificial intelligence (AI) into nuclear systems engineering to improve efficiency and compliance. The Generative Systems Engineering (GenSE) project is transforming traditional systems engineering processes by leveraging AI across the entire plant lifecycle. Key challenges addressed include the extraction and reformulation of requirements, their allocation within the Product Breakdown Structure (PBS), and integration with existing engineering tools. To meet these challenges, a specialized Large Language Model (LLM) tailored for Nuclear Engineering, named "CurieLM", has been developed through fine-tuning. A workflow has been developed, using CurieLM, to automate requirements extraction, ensure quality assurance according to INCOSE guidelines, and facilitate allocation while maintaining compliance with ISO 15288 and ISO 24641 and integrating with SysML tools. The case study on a MOX fuel fabrication plant shows significant time reductions: 88% in requirements extraction, 87% in reformulation, and 66% in allocation to PBS. These improvements are accompanied by a gain in quality, based on feedback from requirements engineers. However, human verification remains essential to interpret and validate the results. In conclusion, the article highlights the potential of AI to transform systems engineering, while highlighting the need for collaboration between humans and AI to guarantee the quality of decisions.
Download

Paper Nr: 14
Title:

AI-Integrated Framework for Enhancing High Level Architecture Design Across System Lifecycle Stages

Authors:

Tianxiao Xu, Néjib Moalla, Mohand Lounès Bentaha, Hazal Aktekin and Claudia Agostinelli

Abstract: AI technology is increasingly being introduced into the automotive industry to support the product design process and address the challenges arising from growing product complexity. Systems Engineering is an interdisciplinary approach and methodology aimed at designing, developing, and managing complex systems throughout entire system lifecycle. The development of Model-Based Systems Engineering (MBSE) significantly enhances complexity management and requirement traceability in conceptual design phases. In the design and analysis phases, the use of Multidisciplinary Design Analysis and Optimization (MDAO) effectively addresses challenges in multidisciplinary problems, identifies optimal solutions, and supports decision-making. Digital Twin (DT) technology is extensively studied and applied to monitor, analyse, and predict operational system behaviour. Integrating AI into system design, along with its combination with MBSE, MDAO, and DT technologies, not only addresses design challenges but also creates new opportunities to advance systems engineering. This paper focuses on how high-level architecture design supports different stages of system lifecycle. The study explores the roles AI can play in the process, as well as its integration with related technologies, and proposes an AI-integrated framework to ensure digital continuity throughout system lifecycle stages.
Download

Paper Nr: 15
Title:

Reinventing Low-Code: Value-Driven and Learning-Oriented Low-Code Development with SLLM-Integrated Approach

Authors:

Gayane Sedrakyan, Stephan Braams, Cosmin Ghiauru, Anton Tsankov, Stijn Schuurman, Matthijs Jansen op de Haar, Valeri Andreev and Jos van Hillegersberg

Abstract: Low-code development platforms (LCDPs) are transforming business practices by shifting the focus from traditional, code-intensive approaches to business-centered modeling. These platforms enable citizen developers - non-technical employees within organizations - to build and manage applications that address specific business needs. This democratization accelerates time-to-market and encourages agile, co-participatory development. However, the rise of citizen development also introduces challenges, such as risks to quality, security, and governance, due to limited technical expertise among some users. This paper investigates ways to enhance current low-code practices by integrating AI-based support for text-to-model generation and established business frameworks, such as the Business Model Canvas (BMC). Incorporating BMC into low-code platforms reinforces their core strengths - minimizing code dependency while grounding development in business models. This integration can offer a structured pathway for citizen developers to engage in meaningful learning while ensuring their projects align with organizational objectives. This approach positions low-code not only as a productivity tool aiming faster time to market, but as platforms for continuous learning and strategic alignment with business. The proposed integrations build on a novel feedback-inclusive approach, which received the innovative feedback nomination at the University of Leuven, Belgium1, and was informed by evidence-based learning experiences at the University of Twente, Netherlands.
Download

Short Papers
Paper Nr: 5
Title:

Multi-Agent Causal Reinforcement Learning

Authors:

André Meyer-Vitali

Abstract: It has become clear that mere correlations extracted from data through statistical processes are insufficient to give insight into the causal relationships inherent in them. Causal models support the necessary understanding of these relationships to make transparent and robust decisions. In a distributed setting, the causal models that are shared between agents improve their coordination and collaboration. They learn individually and from each other to optimise a system’s behaviour. We propose a combination of causal models and multi-agent reinforcement learning to create reliable and trustworthy AI systems. This combination strengthens the modelling and reasoning of agents that communicate and collaborate using shared causal insights. A comprehensive method for applying and integrating these aspects is being developed.
Download

Paper Nr: 6
Title:

Towards an Ontological Approach for Decision Making in Agent Based Systems

Authors:

Sangeeth Saagar Ponnusamy, Philipp Helle and Carsten Strobel

Abstract: One of the key challenges in the design and implementation of decision making techniques in agent based systems is the lack of rigorous system engineering approach between different stakeholders in the development process. This paper proposes a Model Based System Engineering(MBSE) approach based on ontologies to systematically and semantically capture and implement Reinforcement Learning(RL) based decision making in Agent Based Modeling(ABM).

Paper Nr: 8
Title:

Automating Feature Modeling in Product Line Engineering for Systems Engineering: The Application of Natural Language Processing

Authors:

José Lameh, Alexandra Dubray and Marija Jankovic

Abstract: This paper explores the integration of Artificial Intelligence (AI), particularly Natural Language Processing (NLP), with feature modeling (FM) in Product Line Engineering (PLE) for Systems Engineering. By leveraging AI to formalize and model variability, the study proposes an algorithm to assist subsystem owners in describing variability, generating prompts, and producing feature models. The results demonstrate AI’s ability to detect and resolve common modeling issues, such as dead features, false optional features, and constraint inconsistencies, while enhancing model validation and anomaly detection. Although the approach is promising, limitations in scalability, conflict resolution, and integration across subsystems highlight the need for future research to establish a comprehensive and scalable methodology. This work underscores AI's potential to streamline feature modeling and improve the consistency and efficiency of variability management in complex systems.
Download

Paper Nr: 10
Title:

From ML2 to ML2+: Integrating Time Series Forecasting in Model-Driven Engineering of Smart IoT Applications

Authors:

Zahra Mardani Korani, Moharram Challenger, Armin Moin, João Carlos Ferreira, Alberto Rodrigues da Silva, Gonçalo Vitorino Jesus, Elsa Lourenço Alves and Ricardo Correia

Abstract: Time-series forecasting is essential for anomaly detection, predictive maintenance, and real-time optimization in IoT environments, where sensor data is sequential. However, most model-driven engineering (MDE) frameworks lack specialized mechanisms to capture temporal dependencies, restricting the creation of intelligent and adaptive IoT systems. IoT inherently involves sequential data, yet most frameworks do not support time-series forecasting, essential for real-world systems. This paper presents ML2+, an enhanced version of the ML-Quadrat framework that integrates software engineering (SE) with machine learning (ML) in model-driven engineering. ML2+ allows users to define models, things, and messages for time-series forecasting. We evaluated ML2+ through two IoT use cases, focusing on development time, performance metrics, and lines of code (LOC). Results show that ML2+ maintains prediction accuracy similar to manual coding while significantly reducing development time by automating tedious tasks for developers. By automating feature engineering, model training, and evaluation for time-series data, ML2+ streamlines forecasting, improving scalability. ML2+ supports various forecasting models, including deep learning, statistical, and hybrid models. It offers preprocessing capabilities such as handling missing data, creating lagged features, and detecting data seasonality. The tool automatically generates code for time-series forecasting, making it easier for developers to train and deploy ML models without coding.
Download

Paper Nr: 11
Title:

RITSA: Toward a Retrieval-Augmented Generation System for Intelligent Transportation Systems Architecture

Authors:

Afef Awadid, André Meyer-Vitali, Dominik Vereno and Maxence Gagnant

Abstract: Intelligent Transportation Systems (ITS) have significantly transformed the transportation domain by addressing critical challenges such as traffic safety, cost, and energy efficiency. However, the increasing complexity of ITS—arising from the extensive range of applications and technologies they encompass—has made their architectural design modeling time-consuming and challenging, particularly for modelers lacking specialized expertise. Recent advancements in the literature suggest that large language model (LLM)-based modeling assistants offer a promising solution to mitigate these challenges. In this context, this paper introduces the RAG for Intelligent Transportation Systems Architecture (RITSA) project, which seeks to develop a retrieval-augmented generation (RAG) system to support ITS designers/ modelers throughout the architecture design process.
Download

Paper Nr: 13
Title:

Iterative Diagnosis-Driven Augmented Generation (IDDAG) for Programmatic 3D CAD

Authors:

Thomas Paviot, Virginie Fortineau and Samir Lamouri

Abstract: This paper presents a novel approach for automated generation of 3D CAD models using Large Language Models (LLMs) within Model-Based Systems Engineering workflows. We introduce Iterative Diagnosis-Driven Augmented Generation (IDDAG), a methodology combining programmatic geometry creation with systematic diagnostic feedback. The approach leverages a dedicated API for exact Boundary Representation (B-Rep) geometry generation, augmented by a closed-loop architecture that provides iterative refinement based on syntactic, runtime, and geometric analysis. Unlike existing methods requiring extensive training datasets or producing approximate geometries, our solution generates topologically valid, parameterized models while maintaining traceability to engineering requirements. Results demonstrate progressive geometric refinement across iterations, with the diagnostic feedback mechanism effectively identifying and correcting topological inconsistencies.
Download