Application of improved PSO to attack–defense decision models in power systems

Abstract With the development of networked and intelligent power system, the risk of malicious attacks it faces is becoming increasingly severe. Constructing an efficient attack-defense decision-making model is very important to ensure the safe and stable operation of the system. To address the curse of dimensionality and the premature convergence of conventional PSO in large-scale attack–defense problems, this paper deeply studies the application of particle swarm optimization (PSO) algorithm. By designing a particle coding scheme, an adaptive inertial weight adjustment strategy and a multi-objective processing mechanism that fit the characteristics of attack-defense games, the performance of the algorithm in searching for the optimal allocation scheme of defense resources and minimizing the system loss load is significantly improved. Simulation experiments show that compared with the classical optimization algorithm, the improved PSO algorithm shows better global optimization ability, faster convergence speed and stronger robustness when solving large-scale power system attack and defense models, and can effectively support Decision makers formulate fast and accurate defense strategies to improve the resilience of power systems in confrontational environments.