CAJ | 학술논문

粒子群优化一直存在过早收敛的问题。为此，提出一种自动重组机制，当检测到粒子群过早收敛时，会自动触发各粒子的位置重组，从而解放陷入局部最优的粒子群，使其能回到搜索全局最优的正轨。另外，粒子群重组后的新搜索空间的建立，会根据粒子位置与全局最优位置之间偏差的不确定性程度，按照一定比例在各维度上取得新的搜索范围。实验结果表明，在运用于二维 Rastrigin 函数的基准测试中，该优化算法取得了近似全局最优解。
입자군우화일직존재과조수렴적문제。위차，제출일충자동중조궤제，당검측도입자군과조수렴시，회자동촉발각입자적위치중조，종이해방함입국부최우적입자군，사기능회도수색전국최우적정궤。령외，입자군중조후적신수색공간적건립，회근거입자위치여전국최우위치지간편차적불학정성정도，안조일정비례재각유도상취득신적수색범위。실험결과표명，재운용우이유 Rastrigin 함수적기준측시중，해우화산법취득료근사전국최우해。
Aiming at premature convergence of particle swarm optimization,a kind of automatic reorganization mechanism was proposed.When detecting premature convergence of particle swarm,it was able to automatically trigger position reor-ganization and consequently liberate particle swarm in partial optimization and make it back to the right path of searching for global optimum.In addition,establishment of the new searching space after particle swarm reorganization would be on the basis of uncertainty degree of deviation between position of the particle and global optimum position.Meanwhile,ac-cording to a certain proportion,a new searching range was able to be acquired.Experimental results indicated that in bench-mark testing applying two-dimension Rastrigin function,this optimization algorithm was able to acquire approximate global optimum solution.