AI-based bridge generator
Table of contents
Projektdaten
|
Titel | Title |
AI bridge design in HyBridGen
Short description
Bridges are designed by engineers based on their knowledge and individual experiences. This can lead to variations in design quality. Additionally, time-consuming steps are necessary before citizens can be involved in decisions on infrastructure measures. Therefore, AI applications are to be used to generate high-quality and constructible designs efficiently using existing route and geodata, and to design bridges that better meet the needs of citizens.
In the project HyBridGen, a largely automated bridge generator will be developed and tested through a specific example project. In this project, we will use AI to generate plausible bridge designs in an early planning phase by processing BMDV datasets along with engineering knowledge and experiences, as well as project-specific boundary conditions. These designs will be visualized photorealistically and explained in detail. This project will serve as a prototype to enable human-machine interaction for complex infrastructure projects with reduced risks.
An ontology is built up from interdisciplinary requirements, which contains knowledge and information about bridges and their boundary conditions. This ontology will serve as the database for the AI application. Based on design criteria (environmental data, load capacity, budget, etc.), machine learning is used to derive a fast, valid design that is transferred to the generator using parameterized input. The HyBridGen will be validated in a pilot project in the Central German lignite mining region (Saxony-Anhalt).