Numerical modelling enables faster train weighing without an expensive infrastructure

Partner: Drážní revize s.r.o.

Field: railway industry

Drážní revize s.r.o. decided to test the feasibility of the ambitious goal of developing a new train weighing system to overcome existing systems’ technical limitations. This is because traditional rail scales are extremely difficult to build and require massive structures (e.g. deep concrete foundation tubs) that reduce the environmental effects on the measurement. In addition, measurements using these scales are made at very low speeds, and standard passage through the site where the scales are located is only allowed at limited speeds.

Therefore, the representatives of Drážní revize were looking for a solution that would allow trains to be weighed at higher cruising speeds and, more importantly, could be used anywhere without the need to build expensive static concrete structures and infrastructure.

Thanks to the services of the National Competence Centre in HPC, they connected with experts from IT4Innovations National Supercomputing Center. Together, they began developing an innovative concept that uses an advanced combination of artificial intelligence methods with numerical modelling and simulation methods.

This new approach, where data obtained from sensor measurements are supplemented with data generated by numerical modelling and simulation methods, will provide a lot of additional information that cannot be measured experimentally or would be very expensive to measure.

This innovative solution can significantly expand datasets, leading to more accurate and efficient neural network training.

The result can then be a system that not only allows weighing at higher speeds but also does not require complex and costly infrastructure, such as concrete tubs.

The collaboration with Drážní revize shows how combining modern technology, data modelling, and deep learning can deliver revolutionary solutions even in traditionally conservative areas such as train weighing.