President of South Ural State University Aleksandr Shestakov and senior lecturer at the Department of Mathematical Support of Information Technologies of the SUSU School of Electronic Engineering and Computer Science Dmitriy Yaparov have suggested a new approach to processing of dynamic measuring data, which is based on the effect of self-regularization. Their article has been published in the Measurement: Sensors top-ranking international journal, indexed in Q2 of Scopus.
“We need dynamic measurements when we study high-speed processes – for instance, shocks, explosions, or pressure and temperature jumps,” shares Dmitriy Yaparov. “We measure these physical parameters with a device, but due to the time lag in sensors it can yield a big error. In addition, noise can corrupt the device’s readings. If initially the device’s inaccuracy equals just several proportions of one percent, it may turn out to be dozens of percent if we do the dynamic measuring.”
The task is to reconstruct the initial picture using “noisy” data. And here computational mathematics comes to rescue.
SUSU postgraduate Dmitriy Yaparov together with his research advisor Aleksandr Shestakov suggest a new method of data processing based on the effect of self-regularization.
“We must adjust the regularization parameters in such a way that we neutralize the negative effects from noise,” Dmitriy Yaparov explains the method’s idea. “An interval between two measurements can act as such a parameter. And if the system does not allow to perform such regulation, then we introduce the stabilizing functional to the computational scheme of the method of the input signal restoring.”
As a result, it becomes possible to restore data to such a high degree of accuracy that the error in them can be compared to the error of the reference measurement.
Our region’s industrial enterprises have shown their interest in this development, and postgraduate Dmitriy Yaparov (currently working on his thesis) has already obtained two acts of implementation into production: for manometers, and devices for railways.
The research results have not only been published in a prestigious Measurement: Sensors journal, but also reported on at the IMEKO 2024 International Congress in Hamburg. This is the world’s most reputable scientific forum on measurement equipment and metrology.
It might seem that the task has been completed. But Dmitriy Yaparov shares on further prospects. For example, they are planning on using neural networks to process measurement data.
“The results obtained at the interface of control theory and computational mathematics can potentially give start to a new scientific field,” believes SUSU President Aleksandr Shestakov. “There are plenty of tasks for the future research studies, both for Dmitriy Yaparov and for other postgraduates as well.”
The method development has been conducted with the support from the 2024 State Assignment.