The aim of the “KiSiDi” project is to automate the inspection measures for the maintenance of railroad tracks with the help of AI.
The results of the inspection of the rail section are stored on a digital twin in 3D, evaluated and made available to the operator.
The digital 3D representation enables improved collaboration between maintenance and network operators.
In a nutshell
The realization of the project will make the inspection of the rail network more accurate, efficient and cost-effective. Thanks to the provision of digital documentation based on the digital twin, action and maintenance planning and subsequent repair work can be optimized. The result is a unified, interactive and easily accessible data basis for cross-company collaboration. In addition, digital documentation of this kind can simplify and improve the determination of causes following rail traffic accidents.
Modern machine learning methods are used for the AI-based damage analysis. For this purpose, the data stock of already performed rail inspections is processed and labeled. When creating the digital 3D twin, it is necessary to represent the type and position of the damage with a very high degree of accuracy. The measurement results of the different inspection methods are used to achieve this level of detail and to generate an accurate 3D image. Via defined interfaces, the resulting damage analyses as well as the 3D twin shall be made available to network operators and maintenance companies.
Regular and careful maintenance and repair of rail systems is essential to ensure safe and reliable passenger and freight traffic. To date, the evaluation of the rail system has proved to be enormously time-consuming, as the data from existing test procedures must be prepared and analyzed manually. Manual evaluation is associated with a non-negligible susceptibility to error. Thus, the derived results have to be validated subsequently, which results in additional efforts. This leads to inefficient evaluation and the associated maintenance of existing rail networks.
Furthermore, unknown, existing correlations between track geometry and rail defects must be worked out independently. This can mean that safety-relevant deviations are sometimes detected only very late or not at all. This can only be reliably avoided with the help of preventive maintenance.
*with personal injury in 2018
The August-Wilhelm Scheer Institute develops the following components:
In this project, the August-Wilhelm Scheer Institute focuses on the research and development of an innovative approach for the automatic generation of a digital 3D twin from optical and non-optical data of the inspection procedures. The digital twin reflects at any time the condition of an inspected section of roadway and integrates found damages as well as metadata on a three-dimensional model that can be used for planning and rehabilitation procedures and displayed by novel visualization technologies such as AR/VR.
The KiSiDi project is funded by the German Federal Ministry of Education and Research.
Funding code: 01IS21029B
Duration: 01.10.2021 – 30.09.2024