CAJ | 학술논문

传统粒子群算法运行机理是通过粒子群全局最优和自身经验最优来搜索最优位置，不断迭代进化，以此趋近最优解，但该算法共享信息的局限性使其容易陷入局部最优。针对传统粒子群算法的不足，提出了共享历史最优搜索信息的粒子群算法。该粒子群体在搜索过程中，共享算法本次运行的种群个体历史最优信息、当前全局最优信息，及前几次运行过程中的种群个体历史最佳信息。通过5个经典函数的仿真实验测试，验证了该算法具有较强的全局搜索能力和收敛性。
전통입자군산법운행궤리시통과입자군전국최우화자신경험최우래수색최우위치，불단질대진화，이차추근최우해，단해산법공향신식적국한성사기용역함입국부최우。침대전통입자군산법적불족，제출료공향역사최우수색신식적입자군산법。해입자군체재수색과정중，공향산법본차운행적충군개체역사최우신식、당전전국최우신식，급전궤차운행과정중적충군개체역사최가신식。통과5개경전함수적방진실험측시，험증료해산법구유교강적전국수색능력화수렴성。
In view of the limitation of the traditional particle swarm optimization algorithm in sharing information and the defect of being easy to trap the optimized parameter into local optimum,a PSO algorithm which can share historical optimal information was proposed.In the searching process of the algorithm proposed,the group particles of new generation will share the particle historical optimal information of population in current run,the current global optimal information,and the historical individual optimal information of population in previous run.Five classic functions were used to test the new algorithm’s effect,and its stronger global searching ability and faster convergence speed were proved.