Particle swarm optimization algorithm and its parameters: A review

In the year 1995, Dr R.C. Eberhart, who was an electrical engineer, along with Dr. James Kennedy, a social psycologist invented a random optimization technique which a was later named as Particle Swarm Optimization. As the name itself asserts that this method draws inspiration from natural biotic life of swarms of flocks. It uses the same principle to find most optimal solution to problem in search space as birds do find their most suitable place in a flock or insects do in a swarm. The PSO algorithm is initialized with a horde of particles which are a collection of random feasible solutions. Every single particle in the swarm is initialised a random velocity and as soon as they are assigned a velocity these particles start moving in problem search space. Now from this space the algorithm draws the particle to most suited fitness which in turn pulls it to the location of best fitness achieved across the whole horde. The PSO update rule comprises of many distinguishing features which are adjusted and modified depending upon the area of application of algorithm. This paper gives a detailed description of the PSO algorithm and significance of the various parameters involved in its update rule. It also highlights the advantages and disadvantages of using PSO algorithm in any optimization problem.