2.3.2 PARTICLE SWARM OPTIMIZATION (PSO)
Particle Swarm Optimization [27] is a population-based stochastic optimization developed by Dr. Ebehart and Dr. Kennedy in 1995, inspired by social behavior of bird flocking or fish schooling. In PSO, each single solution is a “bird” (particle) in the search space of food (the best solution). All particles have fitness values evaluated by the fitness function (the cost function for ELD problem), and have velocities that direct the “flying” (or evaluation) of the particles. Initialized with a set of random particles (solutions), PSO searches for the optimal solution by updating generations in each iteration. All particles are updated by two “best” values—one called the pbest or personal best …show more content…
When applied to ELD problem, the solution produced by the algorithm is represented by the location of source of nectar while the amount of nectar represents the quality (fitness) of the solution. Employee bees fly around in search of source of nectar (representing trial in a search space) and select their preferred source of nectar based on their experience. Once search is completed, they share their findings (source) with the onlooker bees waiting in the hive. The onlooker bees then make a probabilistic selection of new source of nectar based on the information received from the employee bees. Only if the amount of nectar of the new source is higher than that of the old one, the onlookers choose the new position. If the quality of solution is not improved by a predetermined number of trials (finding sources with higher nectar), then the scout bees fly to choose new source randomly abandoning the old …show more content…
The studies conducted by the researchers confirm that the PSO method itself can be used as an effective and powerful technique for optimizing ELD solution. However, one of its prominent weaknesses found is that it may get stuck into local optima if the global best and local best positions become identical to the particle’s position repeatedly. To alleviate this drawback, hybrid methods combining PSO with other global optimization algorithms like GA, IF, EP, FA, ABC, GSA have been used. Now, in addition to these, a new hybrid of PSO is suggested by the present authors by combining PSO with ACO to study whether better optimization of ELD solution can be achieved. To the best of our knowledge, study of ELD problem solving using PSO-ACO hybrid approach has not been reported in the literature. In this approach, new generation members can be produced at each iteration using PSO and then ACO algorithm can be applied to create extended opportunity of fine-tuning the members.
In PSO algorithm, if the gbest value does not change over few iterations, other particles are drawn closer to the gbest position. As the velocity of the gbest particle gradually reduces by iteration, exploring the local search space (by the best agent) also diminishes. Here the ACO algorithm comes into play. Taking the gbest particles as the input, following the schedule activities of ACO, the ant’s response functions are
Particle Swarm Optimization [27] is a population-based stochastic optimization developed by Dr. Ebehart and Dr. Kennedy in 1995, inspired by social behavior of bird flocking or fish schooling. In PSO, each single solution is a “bird” (particle) in the search space of food (the best solution). All particles have fitness values evaluated by the fitness function (the cost function for ELD problem), and have velocities that direct the “flying” (or evaluation) of the particles. Initialized with a set of random particles (solutions), PSO searches for the optimal solution by updating generations in each iteration. All particles are updated by two “best” values—one called the pbest or personal best …show more content…
When applied to ELD problem, the solution produced by the algorithm is represented by the location of source of nectar while the amount of nectar represents the quality (fitness) of the solution. Employee bees fly around in search of source of nectar (representing trial in a search space) and select their preferred source of nectar based on their experience. Once search is completed, they share their findings (source) with the onlooker bees waiting in the hive. The onlooker bees then make a probabilistic selection of new source of nectar based on the information received from the employee bees. Only if the amount of nectar of the new source is higher than that of the old one, the onlookers choose the new position. If the quality of solution is not improved by a predetermined number of trials (finding sources with higher nectar), then the scout bees fly to choose new source randomly abandoning the old …show more content…
The studies conducted by the researchers confirm that the PSO method itself can be used as an effective and powerful technique for optimizing ELD solution. However, one of its prominent weaknesses found is that it may get stuck into local optima if the global best and local best positions become identical to the particle’s position repeatedly. To alleviate this drawback, hybrid methods combining PSO with other global optimization algorithms like GA, IF, EP, FA, ABC, GSA have been used. Now, in addition to these, a new hybrid of PSO is suggested by the present authors by combining PSO with ACO to study whether better optimization of ELD solution can be achieved. To the best of our knowledge, study of ELD problem solving using PSO-ACO hybrid approach has not been reported in the literature. In this approach, new generation members can be produced at each iteration using PSO and then ACO algorithm can be applied to create extended opportunity of fine-tuning the members.
In PSO algorithm, if the gbest value does not change over few iterations, other particles are drawn closer to the gbest position. As the velocity of the gbest particle gradually reduces by iteration, exploring the local search space (by the best agent) also diminishes. Here the ACO algorithm comes into play. Taking the gbest particles as the input, following the schedule activities of ACO, the ant’s response functions are