Model-Based AI-Driven Systems Engineering (MBSE-AI): Towards Fast and Effective Decisions

Model-based Systems Engineering (MBSE) is increasingly popular as the rising of systems complexity. MBSE replaces documents with models as artifacts which can be used as virtual playground for risk-free experiments, enabling early detection of design flaws. Nevertheless, the current MBSE practices often rely on manual design and subjective analysis due to lack of scientific methodology. Lack of standardized language between project partners also hinders the collaborative work. These challenges result in poor design and decision latency, indicating the need to advance MBSE. This study proposes a Model-based AI-driven Systems Engineering (MBSE-AI) framework that incorporates Artificial Intelligence (AI) into the MBSE workflow, fastening while ensuring highquality decisions. Three AI techniques are synergized, namely Generative AI for enhancing knowledge management and communication, AI optimizer for optimizing the model accuracy and design via sophisticated search, and Agentic AI for improving model accessibility and customizability with automatic model execution. In addition, a Customizable Modeling approach is proposed for improving the flexibility and scalability of the models, supporting the AI-models integration. Finally, the MBSE-AI application is demonstrated for earlystage development of production system in pharmaceutical manufacturing with discrete-event simulation for high-level conceptualization supported with 3D animation, showing the great potential in enhancing MBSE.