Multi-Mean Scout Particle Swarm Optimization (MMSCPSO) based Reactive Power Optimization in Large-Scale Power Systems

Abstract

The particle swarm optimization (PSO) is a population-based algorithm belonging into metaheuristic algorithms and it has been used since many decades for handling and solving various optimization problems. However, it suffers from premature convergence and it can easily be trapped into local optimum. Therefore, this study presents a new algorithm called multi-mean scout particle swarm optimization (MMSCPSO) which solves reactive power optimization problem in a practical power system. The main objective is to minimize the active power losses in transmission line while satisfying various constraints. Control variables to be adjusted are voltage at all generator buses, transformer tap position and shunt capacitor.  The standard PSO has a better exploitation ability but it has a very poor exploration  ability. Consequently, to maintain the balance between these two abilities during the  search process by helping particles to escape from the local optimum trap, modifications were made where initial population was produced by tent and logistic maps and it was subdividing it into sub-swarms to ensure good distribution of particles within the search space. Beside this, the idle particles (particles unable to improve their personal best) were replaced by insertion of a scout phase inspired from the artificial bee colony in the standard PSO. This algorithm has been applied and tested on IEEE 118-bus system and it has shown a strong performance in terms of active power loss minimization and voltage profile improvement compared to the original PSO Algorithm, whereby the MMSCPSO algorithm reduced the active power losses at 18.681% then the PSO algorithm reduced the active power losses at 15.457%. Hence, the MMSCPSO could be a better solution for reactive power optimization in large-scale power systems.