西南师范大学学报（自然科学版）
西南師範大學學報（自然科學版）
서남사범대학학보（자연과학판）
JOURNAL OF SOUTHWEST CHINA NORMAL UNIVERSITY
2014年
1期
47-54
,共8页
粒子群优化%带经验交流的粒子群算法%经验共享%稳定性分析
粒子群優化%帶經驗交流的粒子群算法%經驗共享%穩定性分析
입자군우화%대경험교류적입자군산법%경험공향%은정성분석
particle swarm optimization(PSO)%particle swarm algorithm with communication of experi-ence(PSOCE)%experience sharing%stability analysis
从社会学的角度出发提出了带经验交流的粒子群优化算法(PSOCE)，以克服标准粒子群算法(PSO)在对高维度、多极值函数寻优时收敛缓慢、精度不足、易早熟以及成功率低等问题。算法将粒子个体经学习和积累所得到的经验交予群体社会共享，使每一个粒子个体在学习和经验积累的过程中能够借鉴其他粒子个体已经取得的成果或结论，将经验的效用最大化。利用经典离散控制理论分析其定值算法的稳定范围。仿真分析证明，针对高维度、多极值的目标函数，所提新算法较标准粒子群算法在收敛速度、寻优精度、成功率以及期望迭代次数等方面都有大幅改善。
從社會學的角度齣髮提齣瞭帶經驗交流的粒子群優化算法(PSOCE)，以剋服標準粒子群算法(PSO)在對高維度、多極值函數尋優時收斂緩慢、精度不足、易早熟以及成功率低等問題。算法將粒子箇體經學習和積纍所得到的經驗交予群體社會共享，使每一箇粒子箇體在學習和經驗積纍的過程中能夠藉鑒其他粒子箇體已經取得的成果或結論，將經驗的效用最大化。利用經典離散控製理論分析其定值算法的穩定範圍。倣真分析證明，針對高維度、多極值的目標函數，所提新算法較標準粒子群算法在收斂速度、尋優精度、成功率以及期望迭代次數等方麵都有大幅改善。
종사회학적각도출발제출료대경험교류적입자군우화산법(PSOCE)，이극복표준입자군산법(PSO)재대고유도、다겁치함수심우시수렴완만、정도불족、역조숙이급성공솔저등문제。산법장입자개체경학습화적루소득도적경험교여군체사회공향，사매일개입자개체재학습화경험적루적과정중능구차감기타입자개체이경취득적성과혹결론，장경험적효용최대화。이용경전리산공제이론분석기정치산법적은정범위。방진분석증명，침대고유도、다겁치적목표함수，소제신산법교표준입자군산법재수렴속도、심우정도、성공솔이급기망질대차수등방면도유대폭개선。
On a sociology point of view,a particle swarm optimization algorithm with communication of ex-perience(PSOCE)has been presented for the problem that the standard particle swarm optimization(PSO) has a low convergence speed,insufficient accuracy,and low success ratio,and is easy to get local opti-mums when it is applied to the multidimensional and multimodal functions.This new algorithm shares the experience of each particles with the whole society,and make each particles reference the achievements and conclusions obtained by the others during the process of learning and experience accumulation,in order to maximize the utility of experience.Stability region of its deterministic version in a dynamic environment has been analyzed by means of the classic discrete control theory.The simulation analysis shows that,in terms of the multidimensional and multimodal functions,the new algorithm has a substantial improvement in convergence speed,accuracy,success ratio and expectation of iterations relative to the standard particle swarm optimization.
