16:22, 16 January 2026

Russian Scientists Develop Smart Voltage Control System

The technology addresses a core challenge facing modern electronics.

Researchers at Perm Polytechnic University have developed an intelligent control system for power supplies that uses a neural network to stabilize voltage even when the input source is unstable. The technology tackles one of the key problems in modern electronics powered by solar panels, batteries, and other autonomous energy sources.

DC-DC Converter With an Adaptive Neural Network

Voltage fluctuations, which are inevitable when working with renewable energy sources, can damage equipment. The new system developed in Perm ensures stable output, a critical requirement for the growth of autonomous energy systems, robotics, and portable electronics.

As the university’s press service told IT RUSSIA, the core of the development is a DC-DC converter — a universal “translator” of electrical energy — equipped with an adaptive neural network instead of conventional control circuits. The system accounts for nonlinearities and inaccuracies that inevitably arise in real electronic components. Over time, it adapts to specific operating conditions, compensates for measurement errors and changes in component characteristics caused by heating or aging. As a result, output voltage becomes more stable and regulation smoother.

“One key advantage of our system is that it uses adaptive control algorithms. In real time, the microcontroller analyzes the converter’s output voltage, compares it with the target value, and automatically adjusts the operating mode,” said Vyacheslav Nikulin, associate professor in the university’s Department of Automation and Telemechanics.

Stabilization Error Reduced to Near Zero

The system was tested through both computer simulations and real-world experiments.

“To demonstrate effectiveness, we conducted comparisons in a simulation environment as well as through full-scale physical testing. Both algorithms were evaluated under identical conditions that replicate operation from an unstable source, such as a solar panel,” Nikulin added.

The results were striking. Stabilization error was reduced to near zero. Response time to changes dropped by 58%, from 125 to 79 microseconds. In addition, the system requires 2.5 times fewer computational operations to achieve stabilization — 24 compared with 57.

The technology is expected to serve as a foundation for next-generation power supplies capable of ensuring uninterrupted operation of telecommunications systems.

Energy and housing and communal services