Application of programming methods for control of fluctuations in electric power systems and networks

The purpose of this study was to increase the stability of electric power systems to fluctuations through the development and verification of control algorithms based on advanced programming and modelling methods. To solve this problem, algorithms based on proportional-integral-derivative (PID) regulators (including those optimised by evolutionary methods) and artificial neural networks were created and tested. Tests have shown that the classic PID controller can reduce the amplitude of vibrations by an average of 20-30% compared to an uncontrolled system, however, it required fine manual adjustment and was inferior in response speed to sudden load changes. Optimised PID controllers based on genetic algorithms (GA), particle swarm optimisation (PSO), and the firefly algorithm (FA) helped to further reduce the oscillation amplitude (up to 25%, 33%, and 45%, respectively) and accelerated system stabilisation, which significantly increased the reliability of power supply. Of particular interest were neural networks that provided the highest adaptability to changing conditions and allowed predicting changes in key parameters (frequency and voltage) with an error of 2-3% in the Mean Absolute Percentage Error (MAPE) indicator. As a result, the network responded to disturbances in a timely manner, reduced the frequency deviation to 0.09 Hz, and reduced the transition time to 3.5 seconds in case of sudden load changes. Thus, the neural network approach has demonstrated the best results in both vibration damping and overall stability of the system. The conducted pilot tests in conditions of intelligent power systems have confirmed the feasibility of integrating the developed algorithms into existing monitoring and control infrastructures. With sufficient computing power and an advanced telemetry system, all the proposed solutions were easily scalable and provided reliable vibration damping even in conditions of active integration of renewable energy sources. Thus, the results of the study confirmed the effectiveness of the developed oscillation control methods and their prospects for further widespread implementation in intelligent power systems