SO UES will involve AI in ensuring the quality of data in the energy system of Russia

As TAdviser discovered, the "Unified Energy System System Operator" ("UES SO") decided to study and test the capabilities of artificial intelligence (AI) in processes related to the quality of synchronized vector measurement (VSI) data. On December 8, 2025, the company announced a competition for the implementation of research and development on the topic "Determination of optimal methods and development of algorithms for analyzing the causes of losses and delays in the delivery of IIS to the main dispatch center" SO UES "based on neural networks"^[1].

To begin with, briefly about what synchronized vector dimensions are and what they are for. In order to manage such a complex and geographically distributed object as the electric power system of Russia as a whole, high-precision parameters of the electric mode at various points are required, synchronized in time with global satellite navigation systems and transmitted at high frequency. To solve this problem, the IEEE C37.118 standard was created, which describes synchronized vector dimensions. It follows that the IIS is data measured at the same time with respect to a reference signal common to all measurement points.

𝓲

Фото: Rawpixel.com/Freepik

The terms of reference for the tender "SO UES," published in the open system of the UIS "Procurement," indicate that these IIS are used in the tasks of operational dispatch management of the operating modes of the unified energy system of Russia. To collect SIS data, there is a special automated system operating in the dispatch centers of SO UES, which collects SIS data from more than 180 electric power facilities - from more than 1150 devices of synchronized vector measurements (USVI) with sampling 50 times per second. The volume of information processed reaches 25 million television measurements per minute.

The introduction of information and control systems in the dispatching centers of SO UES JSC, operating on the database of IIS in real time, poses very high requirements for the quality of data flows. An increase in the level of losses and delivery time of online data of the IIS reduces the reliability of the monitoring information systems and leads to the provision of irrelevant or false information to the dispatching personnel, the terms of reference says. And when it comes to the use of data from hundreds of stations and thousands of devices of the transient monitoring system (SMPR), special monitoring tools are needed.

In SO UES, data quality monitoring software, which determines the delivery time of television measurements and loss levels for each channel, was put into operation in 2020. It is used by control center technologists to identify the facts and causes of degradation of IIS data.

However, recently there has been a significant increase in the number of cases of data quality degradation due to the suppression of satellite synchronization signals, while due to different methods of synchronizing measurements, significant differences in the response of devices and CMCP complexes on objects are recorded when suppressing satellite time signals. This fact, as well as the presence of many nodes that interfere with each other on the information transmission path, revealed the need to develop analysis tools based on AI, since manual diagnosis of anomalies due to a large number of elements (more than 10,000) becomes practically unrealizable, - explains "SO UES" in the published terms of reference.

In the IDR that the company wants to conduct, studies should be carried out to determine the optimal methods for analyzing the causes of losses and delays in the delivery of IIS data to the main dispatch center based on neural networks with the RNN/LSTM/Transformer architecture. It is also necessary to develop algorithms that ensure the fixation of the causes of the corresponding data degradation, implement and test them in a test module connected to the streams of real data of the IIS in the main control center.

The purpose of the work is to improve the process of monitoring the quality of IIS data when they are collected in the dispatch center from devices and complexes of SMPR installed at electric power facilities.

As the customer explains in the terms of reference, the use of AI in this case is an effective solution primarily due to the fact that the dynamics of degradation of the quality of traffic of data IIS in various cases is reflected in different ways in the measured metrics, such as the amount of time delay, the level of loss, the number of personnel gaps, etc., which makes it very difficult to solve this problem using a conventional mathematical apparatus.

It plans to select an executor for the research and development "SO UES" in mid-January 2026.

Notes