Multiobjective reactive power compensation with an ant colony optimization algorithm

Abstract

This paper presents an Ant Colony Optimization (ACO) algorithm applied to the reactive power compensation problem in a multiobjective context. The developed algorithm was denominated Electric Omicron (EO) given that it was inspired in the Omicron ACO proposed by some of the authors. The proposed EO algorithm was compared to a variant of the SPEA (Strength Pareto Evolutionary Algorithm), specially designed for this problem. This variant of SPEA has previously shown an excellent performance in this type of problem. Experimental results presented in this paper show that the proposed EO outperforms SPEA, i.e., EO finds better Pareto solutions considering voltage deviation and investment. As long as we know, this is the first attempt to solve the reactive power compensation problem with an ACO algorithm in a multiobjective context.

Multiobjective Optimization in Optical Networks

“…It is also possible to implement other approaches that use more than one pheromone matrix (Iredi et al, 2001;Doerner et al, 2002;Gardel et al, 2006). Note that the hop-count visibility matrix d ij is an input data of the problem.…”

NSGA-II with local search for a multi-objective reactive power compensation problem

“…Finally, the apparent power equations at bus m can be written since the apparent power that enters bus m must be equal to the sum of all power that goes out of it, (15).…”

“…For the first iteration losses are considered null and all the bus voltages are 1 p.u. The apparent power reaching any bus is the sum of all apparent power that is possible to distribute from this bus to the remaining network (15). The power entering the first bus, the substation, is the total apparent power necessary for the normal operation of the network.…”

“…More recently, heuristics [11] and mainly meta-heuristics, such as Simulated Annealing [12][13][14], Ant Colony [15], Particle Swarm Optimization [16], Plant Growth Simulation Algorithm [17] and Tabu Search [18], Fuzzy based techniques [19] and Genetic Algorithms [20], have been used to deal with the problem. Because of the characteristics of the VAR planning problem, which is a combinatorial non-linear multi-objective mixed integer problem, metaheuristics can cope with model complexity and tractability, reduce the exhaustive search in large spaces and lead to (near) Pareto optimal solutions.…”

Genetic algorithm approach in FACTS devices location for the improvement of energy efficiency in distribution networks

“…Swarm Intelligence applied to power systems includes Ant Colonies Optimization (ACO), where artificial ants build solutions by moving on the problem graph and changing it so that future ants are capable of building better solutions. Problems such as losses minimization, reactive power compensation and system restoration have been solved using this kind of approach, as presented in [1,2,3,4,5]. In the same approach area lays Particle Swarm Optimization (PSO).…”