A comparative study of performance in particle swarm optimization methods with reflection

In this paper, two kinds of Reflectance-Adjusting PSO (RAPSO) methods that improve the adjustment of the reflectance in the vector reflection PSO are proposed. One is RAPSO using the standard deviation of optimal evaluation values for the latest steps, called RAPSO-OE. The other is RAPSO using the standard deviation of particles´ evaluation values in the present step, called RAPSO-PE. The proposed methods are compared with Simple PSO and Taper-off-Reflectance PSO (TRPSO). The validity is shown by benchmark tests, and the proposed method is applied to the optimization problem of electric power leveling systems in rolling mills. The simulation result shows that adjusting reflectance is effective for reducing search time, especially when the optimal solution exists in the edge of problem domain.