Study of the Efficiency of Optimizing the States of Complex Power Systems Using Artificial Intelligence Methods

Abstract

Relevance. at present, a steady increase in global demand for electrical energy, the limited availability
of conventional resources, and the necessity to ensure environmental safety and sustainability require
more efficient use of resources in the energy production process. Taking these considerations into
account, this paper presents the results of a study on the applicability and comparative performance of
five artificial intelligence–based methods and algorithms for power system state optimization, namely
the Genetic Algorithm (GA), Particle Swarm Optimization (PSO), Differential Evolution (DE),
Artificial Bee Colony (ABC), and Ant Colony Optimization (ACO). This approach is significant not
only from a theoretical perspective but also in terms of practical implementation and enhancing the
stability of power systems.
Aim: the objective of this study is to improve models and algorithms for short-term planning and
operational control of modern complex power systems based on an analysis of the effectiveness of
artificial intelligence methods in optimizing their operating states.
Methods: the research employed modern theoretical approaches to the calculation and optimization of
steady-state operating conditions of electric power systems, as well as nonlinear mathematical
programming, system analysis, and artificial intelligence methods.
Results: in general, the problem of optimizing power system operating states represents a complex
multi-extremal problem involving a large number of equality and inequality constraints of simple,
functional, and integral types. In certain cases, the use of traditional solution methods leads to the
necessity of approximation, which results in reduced accuracy and difficulties associated with the unreliability of iterative computational processes. To overcome these difficulties, the application of
artificial intelligence methods is considered essential. A comparative analysis of their performance was
conducted using a representative case study. The results demonstrate that the Particle Swarm
Optimization (PSO), Differential Evolution (DE), Artificial Bee Colony (ABC), and Ant Colony
Optimization (ACO) algorithms exhibit approximately similar performance and show higher efficiency
compared to the other considered methods.