PARTICLE SWARM OPTIMIZATION

In the year 1995, Kennedy and Eberhart proposed the particle swarm optimization (PSO) technique. This algorithm is motivated by the common behavior of bird groups for food searching. In PSO, the members without mass and without volume depending on the velocities and accelerations in the direction of the best mode of behavior are called particles. Each particle within the swarm represents a solution in the specified high-​dimensional space. It considers four vectors which are its current best positions at the moment. Depending on its neighborhood and its velocity, the best position is found, then based on the best position its position is altered in the search space achieved by itself (p_best) and on the best position achieved by its neighborhood (g_best) during the search process. PSO can be described by a sequence of steps. First, allocate a swarming random location in the searching space and compute the fitness function of each particle. Individual fitness value of particle is compared with the best fitness value, and if the current fitness value is better than the best fitness value, then this value is set as the best solution (x_i as p_i) and the current particle location. The particle is identified with the best fitness value and the velocities and locations of the particles are changed using the same procedure repeated until final requirements are met. PSO is easier to implement, efficient in maintaining the diversity and it has a more memory capability than genetic algorithms.