The Reactive Power Optimization of Distribution Network Based on an Improved Genetic Algorithm

Abstract

The genetic algorithm is the general purpose optimization technique on the principles inspired from biological evolution. In this paper, an improved genetic algorithm is introduced to study the reactive power optimization of distribution network. An improved genetic algorithm is improved in these processes mentioned below: a dynamic retribution factor is employed in the fitness function; a decimal coding method is used for the adjustable transformer ratio and parallel capacitors; the initial population is customized to spread in the whole solution space; a competition algorithm is used for selection to avoid 'premature'; dynamic crossover factor and dynamic mutation factor are employed to increase convergence accuracy and speed; mutation is carried out near the current value to satisfy the device restriction. The distribution network case study shows that the algorithm has a good performance in convergence speed and global optimization