The Use of Large Language Models to Automate the Creation of UML Diagrams for Industrial Systems

Abstract

This article examines the problem of documenting complex distributed systems in the context of the Industry 4.0 concept and the Industrial Internet of Things (IIoT). Traditional methods of graphical modelling are time-consuming and cannot keep pace with the rapidly changing architecture of industrial complexes, which include sensors, gateways, CNC controllers and AI models. The aim of this work is to develop a software solution for the automatic generation of technical documentation based on the 'Diagrams-as-Code' paradigm. A desktop application is presented that utilises large language models to convert textual descriptions into PlantUML and Mermaid diagrams. The results of testing the tool in specific industrial scenarios are presented: designing the architecture of an IIoT enterprise, modelling anomalies in CNC machine tools, describing finite-state machines of controllers, and planning maintenance and repairs. The practical value of this development lies in reducing the time taken to create architectural diagrams and schematics whilst maintaining high-quality output. The proposed tool has the potential to become a key component of a modern enterprise's digital stack, facilitating interdisciplinary collaboration and the integration of documentation into CI/CD.