As part of the "ESPEK" project, the frequency, impact and relevance of certain types of damage to common wagon types were analyzed. Based on this, the research team has developed practical, data-based, AI-supported solutions to automatically detect damage to selected passenger wagons as they pass by.

Since the start of the project in August 2023, the "Safe Processes and Systems" research group headed by Prof. Gillert and the Image Processing department headed by Prof. Alexander Stolpmann at the TH Wildau have been working intensively on the development of AI-supported computer-visual techniques together with the project partners RWS Railway Service GmbH from Wustermark and Telco Tech GmbH from Teltow. Various scientific approaches were tested and a future concept for the digital inspection of passenger trains (PVZ) was developed. The aim of the project was to automate and optimize inspection processes in railroad workshops so that inspectors can carry out maintenance regardless of location.

Automated inspection instead of a "trained eye"

In most workshops, PVZs are still inspected manually and almost exclusively with a trained eye. This is not only time-consuming, but also carries the risk of damage being overlooked. There is also a risk of a loss of knowledge due to the shortage of skilled workers in this sector. The working and environmental conditions in the railroad sector are challenging (e.g. weather, soiling), and an enormous variety of series makes it difficult to develop universal, modular solutions that can be parameterized.

The ESPEK team therefore set itself the goal of examining the current inspection processes in the workshops for their automation potential. The use of state-of-the-art camera technology and artificial intelligence (AI) should optimize inspection work and increase efficiency and safety. Automation could be used for safety-relevant components in particular. Automated inspection enables maintenance measures to be planned at an early stage and resources to be allocated, which can significantly reduce maintenance throughput times.

On September 9, the project partners will present their results in detail. These include the successful prototype implementation of the automated inspection of brake pads and brake discs using 3D camera technology and precisely aligned lighting technology. The finest cracks can be detected and measured during a drive-by, whereby not only sections of the brake discs are examined. The data obtained forms the basis for an overall concept for digital maintenance, which is intended to engage various target groups, standardize future applications in the rail industry and possibly even simplify enterprise content management. This future concept can also have enormous potential in civil security research.

"ESPEK" means: Detection of damage patterns on passenger trains and evaluation of confidence for the selection of robust features for predictive maintenance.