Novosibirsk State Technical University (NETI) is developing a way to monitor the condition of railway tracks. We are talking about creating an accurate digital sample of a complex of engineering structures that will help identify defects on the tracks and eliminate them in time.
NSTU-NETI has proposed an effective way to monitor the condition of railway tracks using the mathematical straightening method. According to Elizaveta Osintseva, a student at the Faculty of Applied Mathematics and Computer Science, it consists in creating an accurate digital passport of railway tracks on specified sections of all types — straight, transitional (smoothly changing direction) and circular (turning).
"Over time, the tracks deform due to operation, temperature fluctuations, loads and other factors. Traditionally, their condition is checked manually: technicians and specialists are used to inspect the sites, which is long and expensive. A new way to monitor the condition of the paths will automate the process. The data collected from the laser scanner (lidar) can be compared with a reference path plan and potential defects can be identified. This will make it possible to find problem areas faster and send specialists only to those where verification is really needed," said Elizaveta Osintseva.
As the student noted, mathematical methods make it possible to automate data analysis and eliminate the need for manual inspection of each meter of the path. It was decided to abandon the use of AI for a while in favor of proven algorithms. "A working group has been established at our faculty under the leadership of Associate Professor of the Department of Theoretical and Applied Informatics, Candidate of Technical Sciences Vitaly Karmanov. In our research, we apply and compare various methods for calculating the curvature of a path, for example, using differential analogs of derivatives, estimating bending arrows or smoothing splines. Some of them are effective on a large amount of data, but are unstable to noise and measurement errors, while others are the opposite. Therefore, it is most effective to use a set of different methods," said Elizaveta Osintseva.
According to the developer, there is no information on analogues of such a system in the domestic market. There are programs for analyzing railway tracks abroad, but they usually do not fully automate data processing. "Moreover, we focus on the full automation of lidar data analysis, which is an innovation even in international practice," said Elizaveta Osintseva.
Earlier, NSTU-NETI started developing software for creating digital models of highways. The customer was the Territorial Administration of Highways of the Novosibirsk region. The development of the program will make it possible to carry out road certification and effectively solve problems related to the development of the automobile transport infrastructure in the Novosibirsk region. The digital model allows you to determine geometric parameters, take into account bumps, stops, road signs and other elements of thousands of kilometers of roads.