- Статьи
- Science and technology
- Faulty power: new algorithms will make solar power plants weather-independent
Faulty power: new algorithms will make solar power plants weather-independent
Russian scientists have developed an algorithm based on artificial intelligence that is able to detect abnormal operating modes of electrical networks and automatically trigger protective mechanisms. After eliminating the threat, the program independently determines the optimal moment to return the equipment to normal operation. The development is particularly in demand in power systems using energy storage and renewable sources. In the future, the technology may become the basis for creating fully automated power grid management systems that do not require human intervention. As experts explained to Izvestia, the effectiveness of the algorithm has so far been confirmed only on computer models. For implementation at real facilities, refinement and testing will be required.
How smart algorithms protect networks
Scientists from the Skolkovo Institute of Science and Technology, the Northeastern Federal University named after M.K. Ammosov and the Center for Practical Artificial Intelligence of the BEAC have developed a smart algorithm that detects and prevents dangerous modes in electrical networks. It also helps to avoid accidents at IT infrastructure facilities and production management systems.
As the researchers explained, the AI fuse continuously monitors the operation of the equipment. If the system detects sudden and chaotic changes in key parameters, the algorithm automatically switches the equipment to safe mode, effectively "freezing" the system. After that, using mathematical models developed by scientists, the program determines the optimal moment for the smooth return of the equipment to normal operation.
— These are not chatbots or computers that "reason" like humans. We are talking about analytical artificial intelligence for analyzing streaming data. It monitors the behavior of networks in real time and detects signs of a transition to abnormal modes," Stepan Vasiliev, one of the developers and a graduate student at the Skoltech Center for Energy Technologies, told Izvestia.
The AI fuse functions as an intelligent sensor, the scientist explained. If the algorithm detects signs of instability — sudden fluctuations, rapid deviations, or atypical dynamics of indicators — it classifies the current mode as potentially dangerous and transmits a signal to the equipment to trigger the stabilizing mechanisms.
After the parameters return to normal values, the program determines that the critical situation has ended, and independently selects the optimal moment to restore normal operation in order to avoid repeated surges and overloads.
According to the expert, the development is particularly in demand in power systems with energy storage, charging stations and renewable energy facilities, including wind and solar power plants, whose operation directly depends on weather conditions.
— Traditional power plants generate energy using heavy turbines, which have a lot of inertia. They sort of "smooth out" sudden changes in the network and extinguish short-term spikes. Renewable sources work through high—speed power electronics, which does not provide such natural inertial support," explained Stepan Vasiliev.
If, for example, a cloud covers the sun or the wind increases sharply, fluctuations in voltage, current and other parameters may occur in the power system, the developer explained. Such processes occur in fractions of a second, but they can lead to equipment shutdowns, accidents, data loss, and even large—scale power outages - blackouts. The proposed method makes it possible to detect dangerous operating modes almost instantly and automatically launch protective scenarios to stabilize the system.
The development can also be in demand in enterprises with a large number of powerful computing and digital equipment — in data centers, banks, stock exchanges, telecommunications hubs and factories with software-controlled machines.
According to Stepan Vasiliev, in the future, the technology will make it possible to create a new generation of control systems capable of operating without constant human involvement and ensuring reliable operation of power grids with a high proportion of renewable generation. In the future, AI fuses may become the basis of intelligent power systems that will independently adapt to changing conditions, including weather factors.
Contribution to global energy security
The practical benefit of the solution lies in the stability of the infrastructure and minimizing downtime of equipment and services, said Nikolai Tiden, Director of the Central Research Institute of the Russian Academy of Sciences.
— This is a transition from passive monitoring to active risk management. Scientists have developed an algorithm that not only fixes failures, but prevents them," he said.
During the study, the approach was tested on energy systems, but the principle itself can be applied much more widely — on any complex objects where critical deviations develop in a short period of time, the expert said.
According to Daniil Sukhorukov, a researcher at the AI in Industry Group at the AIRI Institute, the proposed approach is particularly in demand in microgrids, local power systems where reliability and uninterrupted power supply are critically important. The main advantage of the development is that it is implemented at the software level and does not require the installation of additional expensive equipment.
At the same time, as the expert noted, before introducing the technology into real operation, it is necessary to conduct tests on digital twins, specialized test benches and pilot facilities.
— Green energy facilities are actively being built in our country. Their correct behavior in emergency conditions is a matter of reliability of power supply to entire areas," said Dmitry Averyanov, commercial director of the Linesman installation company.
A large number of semiconductor consumers operate in modern power grids. These include frequency-controlled drives, switching power supplies, LED drivers, and other devices. They are sensitive to voltage fluctuations and can become sources of interference and distortion themselves when the network is unstable. Therefore, in such systems it is especially important that the generation sources work predictably and controllably, he explained.
At the same time, as the specialist emphasized, at the moment the development results are confirmed only on computer models. When switching to real objects, additional difficulties may arise, including delays in transmitting signals between stations and differences in the speed of control systems. This will require further refinement of the algorithms and additional testing.
According to Dmitry Averyanov, similar approaches are being actively studied abroad. The Russian development, he noted, also contributes to solving the global task of creating reliable and safe energy management systems of a new generation.
Переведено сервисом «Яндекс Переводчик»
