Control strategy PSO

• An evaluation index called “Control Strategy PSO” is developed.
• It can be applied to other intelligent algorithms.
• We present a detailed theoretical and empirical analysis.

Many variants of particle swarm optimization (PSO) both enhance the performance of the original method and greatly increase its complexity. Motivated by this fact, we investigate factors that influence the convergence speed and stability of basic PSO without increasing its complexity, from which we develop an evaluation index called “Control Strategy PSO” (CSPSO). The evaluation index is based on the oscillation properties of the transition process in a control system. It provides a method of selection parameters that promote system convergence to the optimal value and thus helps manage the optimization process. In addition, it can be applied to the characteristic analyses and parameter confirmation processes associated with other intelligent algorithms. We present a detailed theoretical and empirical analysis, in which we compare the performance of CSPSO with published results on a suite of well-known benchmark optimization functions including rotated and shifted functions. We used the convergence rates and iteration numbers as metrics to compare simulation data, and thereby demonstrate the effectiveness of our proposed evaluation index. We applied CSPSO to antenna array synthesis, and our experimental results show that it offers high performance in pattern synthesis.

Figure optionsDownload as PowerPoint slide